Weld Size Explained: How to Read, Specify, and Check Welds

A fabricator once shipped a 40-foot girder whose fillet welds looked perfect — until an inspector put a $12 gauge on them and found the legs were a sixteenth short on half the run. That tiny gap between ‘looks big enough’ and ‘is code-compliant’ is exactly where weld size lives, and it decides whether a joint fails in service or simply never gets rejected.

What Weld Size Actually Means (and Why It Is Not One Number)

A fabrication shop once shipped a 40-foot girder whose fillet welds looked textbook perfect: even ripples, clean ties at the toes, minimal spatter. Then an inspector put a standard fillet gauge on them and found the legs were a sixteenth short on half the run. Nothing about those beads would fail an appearance check, but they were not the size the drawing called for, and in structural work that gap is the difference between an accepted member and a rejected one.

That is the space weld size lives in. Before anyone strikes an arc, it pays to pin down exactly what the term means.

The Short Answer: Weld Size Is a Documented Dimension

Weld size is the term drawings and codes use to describe how big a weld is required to be. It is a specified number with rules attached, not a welder’s estimate or a matter of taste. AWS D1.1 states it plainly: full and complete information regarding the location, type, size, and extent of all welds shall be clearly shown on the drawings (AWS D1.1/D1.1M, Structural Welding Code). If the size is not on the print, the weld is not ready to be made.

But “size” does not describe the same thing at every joint, and that is where the confusion starts. The callout on a print looks like a single value, yet that value can mean a leg length, a throat, a depth of preparation, or a plate thickness depending on the geometry in front of you. Same word, different joint, different measurement.

Fillet Weld Size: The Leg of the Largest Inscribed Right Triangle

A fillet weld joins two members meeting at an angle, usually a right-angle T-joint or a lap joint, and its cross section is roughly triangular. Sizing it comes from geometry rather than from the bead you can see: the size is the leg of the largest right triangle that fits inside the weld’s actual cross section, with the hypotenuse lying along the face of the weld. Size is therefore read along the fusion faces, from toe toward root, not across the crown of the bead.

For an equal-leg fillet both legs match and the callout is unambiguous. When the legs are unequal, the inscribed triangle is limited by the shorter side, which is why a fillet that looks generous on one side can still be qualified at the smaller dimension.

This is where welders and inspectors start talking past each other. A welder usually thinks in throat: the distance from root to face, the dimension that actually carries the load. For a true 45-degree fillet the theoretical throat is 0.707 times the leg, so a 1/2-inch leg gives a throat of about 0.354 inch. Prints typically call the leg instead, because the leg is the dimension a fillet gauge can check quickly in the field. Convexity adds real throat beyond that theoretical value, but design calculations use the theoretical throat, so extra reinforcement is extra metal rather than extra strength.

Groove Welds: Size Is the Joint, Not a Triangle

Move to a groove weld and the definition shifts again, because there is no inscribed triangle to measure. For a partial joint penetration groove weld, size is the depth of the joint preparation: a 5/8-inch groove weld on a 5/8-inch bevel. For complete joint penetration, the size effectively becomes the thickness of the thinner member joined, since the weld fills the full depth of the prepared joint. AWS D1.1 treats these as separate cases, and the welding symbol has to carry the difference.

Why One Number Is Never Enough

Notice that the code sentence lists four things together: location, type, size, and extent. Size alone only answers how large the cross section must be. It says nothing about where the weld runs or how long it runs, and a fillet held at a perfect 1/2 inch that stops two inches short of the end of the joint still fails the drawing.

So the honest answer to “what is the weld size?” is that it is a documented dimension tied to a specific joint type, checked against a specific reference geometry. Pin down the joint, the callout, and the measurement method before the arc starts, and a good-looking bead and a compliant weld become the same thing.

The Anatomy of Weld Size: Legs, Throat, and the 0.707 Rule

Before you can size, check, or argue about a weld, you need to see its cross-section clearly. A fillet weld is a roughly triangular fillet of metal dropped into the corner where two members meet, and every dimension that matters comes from that triangle. The confusion starts because the number on the drawing, the number you measure with a gauge, and the number that actually carries load are three different things.

Leg Length and Weld Size

The leg is the distance from the joint root to the weld toe along each of the two joined surfaces. On an equal-leg fillet, both legs are the same length, and that shared length is the weld size. When you see a 1/4 in fillet called out on a drawing, the detailer is specifying a 1/4 in leg on both sides of the weld’s right angle.

Here is the part beginners miss: weld size is defined by the largest right triangle that can be inscribed in the weld face. Not the widest point, not the average, not the meaty middle of a lumpy bead. If a weld looks fat but its profile bulges outward, the inscribed triangle does not grow. That single rule explains why convexity buys you nothing and why a proud-looking weld can still fail inspection.

Three Throats, Not One

The throat is the dimension across the weld that resists the load, and AWS D1.1 draws sharp distinctions between three of them:

• Theoretical throat: the distance from the joint root to the face of the diagrammatic weld, meaning the flat-faced triangle drawn on the symbol. For an equal-leg fillet it is the altitude of that inscribed triangle.
• Actual throat: the shortest distance from the root to the actual weld face, measured on the real bead. On a concave weld this is smaller than the theoretical throat; on a convex weld it is larger.
• Effective throat: the shortest distance from the joint root to the face of the diagrammatic weld. Per AWS D1.1, this is the dimension used for design calculations.

The effective throat is the number that counts. Extra metal piled on top as convex reinforcement is treated as dead weight; it contributes nothing to strength. Concavity, on the other hand, cuts directly into the effective throat and can quietly put a weld below its specified size.

Where 0.707 Comes From

An equal-leg fillet with a flat face forms a 45-45-90 triangle. The throat is the altitude of that triangle, and simple trigonometry gives:

throat = leg × sin 45° = leg × 0.7071

So a 1/4 in leg delivers a theoretical throat of about 0.177 in. A 1/2 in leg delivers 0.354 in. The AISC fillet weld FAQ confirms the design convention: fillet weld stresses are computed on the effective throat and treated as shear, regardless of the direction of the applied load.

For unequal legs, the largest inscribed triangle has two different sides, and the throat becomes (a × b) ÷ √(a² + b²). Design codes generally simplify this conservatively to 0.707 times the smaller leg. If you weld a 5/16 in leg against a 3/16 in leg, the 3/16 in leg governs, and the bigger leg is doing far less than it appears to.

Worked Examples

Leg size	Effective throat (×0.707)	Cross-section area	Metal per inch of length
1/4 in (6.4 mm)	0.177 in	0.031 in²	0.031 in³
3/8 in (9.5 mm)	0.265 in	0.070 in²	0.070 in³
1/2 in (12.7 mm)	0.354 in	0.125 in²	0.125 in³
6 mm	4.24 mm	18 mm²	18 mm³

Notice the jump. Because area scales with the square of the leg, a 1/2 in fillet consumes four times the weld metal of a 1/4 in fillet over the same length, for exactly twice the throat. That square-law penalty is the single most important economic fact in weld sizing, and we will lean on it heavily when we get to distortion and cost.

The geometry is identical whatever the process. A pulsed aluminum MIG fillet in a boat hull obeys the same 0.707 rule as a hardfacing bead on farm equipment, which is why the practical tricks in our aluminum MIG welding guide translate straight across materials.

The takeaway to carry into the next sections: size is a geometric label, throat is the structural reality, and the gap between them is where rejections are born.

How Weld Size Is Specified on Drawings and Welding Symbols

A weld size only exists once it is written down. The engineer calculates it, the detailer puts it on a drawing, the welder deposits it, and the inspector checks it. Everything before and after that moment runs through the AWS welding symbol, so misreading one character can turn a compliant joint into a rejection.

The Basic Fillet Symbol Layout

Every welding symbol hangs off a horizontal reference line with an arrow pointing at the joint. For a fillet weld, the size goes to the left of the reference line, and the length or pitch goes to the right. So a symbol reading 1/4-10 means a 1/4 in leg length in 10 in segments. 1/4-5~10 means 5 in welds at 10 in on-center (the tilde is the pitch mark).

Placement above or below the reference line tells you which side of the joint gets the weld:

• Arrow side (symbol below the line): weld the side the arrow points to.
• Other side (symbol above the line): weld the far side.
• Symbol on both sides: a fillet symbol above and below the line means weld both sides of the joint, and a size listed on only one side applies to both unless stated otherwise.

Two small flags and marks change the meaning again: a flag at the arrow marks a field weld (welded on site, not in the shop), and a box around the entire symbol marks a weld installed around the periphery or a run-on/run-off detail depending on the standard. A break in the arrow points to the shallow side of a bevel or J-groove.

Groove Welds: CJP vs. PJP

Complete joint penetration groove welds are usually called out without dimensions. The AWS D1.1 drafting convention recommends that contract drawings show CJP and PJP groove weld requirements without specifying groove weld dimensions; the shop drawing then specifies groove depth S for the required weld size E (AWS D1.1/D1.1M). The reason is simple: a CJP weld’s strength comes from full penetration through the whole thickness, not from a number in a box. Partial penetration welds are the opposite. They must state both the required effective size E and the groove depth S, because penetration can be shallower than the bevel.

The symbol itself also carries geometry: the bevel, V, U, J, and flare-V shapes tell the welder which groove prep to cut, and a root opening or root face is written beneath the reference line.

Intermittent Welds, End Returns, and Skewed Joints

Intermittent fillets are written as length-pitch to the right of the line, and the minimum length of an intermittent fillet segment is 1.5 in (40 mm) per AWS D1.1. End returns (boxing) show as a short line across the end of the reference line, meaning the weld wraps around the end by a specified distance. When a plate terminates short of a computed length, the drawing may show the weld run out to the physical end instead.

Skewed T-joints are the classic trouble spot. Because the leg no longer sits square to either member, the engineer must specify weld locations and clearly show weld dimensions, including whether effective throat E or leg W controls. A plain fillet symbol here is not enough; the drawing needs a sketch.

While you are reading symbols on an aluminum job, the same rules apply across processes, whether you are working from Aluminum MIG Welding: The Complete Practical Guide or Aluminum Flux Core Welding: What Actually Works in 2026.

Detailer’s Symbol Checklist

1. Confirm the size sits left of the reference line and length/pitch sits right.
2. Check arrow side vs. other side; add symbols on both sides when both are welded.
3. For PJP, give both E and S; for CJP, follow the detailing convention rather than a throat number.
4. Add the field weld flag and check for boxing/end-return marks.
5. Verify intermittent segments meet the 1.5 in minimum length.
6. Sketch skewed joints and state whether E or W controls.
7. Note metric prefixes: a = throat, z = leg.

Symbol literacy is the cheapest quality control on a project. The girder in our opening story failed inspection not because anyone misread a gauge first, but because the drawing left room for interpretation. Related reading on verification habits appears in Aluminum Boat Welding: The Complete Guide for Strong Hull Joints, and process context on joint prep in Aluminum Cleaner for Welding: How to Prep Aluminum Right.

Minimum Weld Size: The Code Floors You Cannot Go Under

Every weld size conversation starts with a floor. Before an engineer calculates anything for load, before a detailer types a number into a symbol, the codes impose a minimum fillet size based purely on how thick the members are. These minimums are not suggestions and they are not strength values. They exist to keep the weld from cracking while it solidifies.

The Minimum Fillet Weld Size Table

AISC 360 Section J2.4 and AWS D1.1 Table 7.7 (Clause 7.13) carry essentially the same numbers:

Thicker part joined (in.)	Minimum fillet weld size (in.)
To 1/4 in. inclusive	1/8
Over 1/4 to 1/2 in.	3/16
Over 1/2 to 3/4 in.	1/4
Over 3/4 in.	5/16

Read it as a lookup: find the thickness of the thicker part, then read across for the smallest leg you are allowed to deposit. A 3/8 in. gusset on a 3/8 in. plate sits in the “over 1/4 to 1/2” row, so 3/16 in. is the floor, even if your calcs say 1/8 in. would carry the load.

Why the Floor Exists: It Is Metallurgy, Not Strength

This is the part people get wrong. A small weld on a thick plate can be perfectly strong enough on paper and still crack in the joint. Thick base metal acts as a heat sink, sucking heat out of a small molten pool so fast that the metal quenches. The rapid cooling creates a hard, brittle heat-affected zone, traps hydrogen, and produces the classic underbead crack that usually appears a few seconds to a few hours after the pass.

The AISC commentary states this explicitly: minimum weld size requirements are not based on strength, they are based on the quench effect of thick material on small welds. The bigger weld takes more heat, stays hot longer, cools more slowly, and gives hydrogen time to escape. That is why the minimum grows with thickness rather than staying flat.

Reading the Footnotes (This Is Where Jobs Get Failed)

The four rows of the table are only half the rule. D1.1’s footnotes change which thickness applies and when:

• Footnote (a) — process matters. With non-low-hydrogen electrodes (like basic E6010 root passes) and no qualified preheat procedure, the thicker part governs, the weld must be single-pass, and preheat is required. With low-hydrogen processes (FCAW, GMAW with shielding gas, SAW, E7018), the thinner part governs and multi-pass welding is permitted.
• Footnote (b) — the cap. No minimum exceeds the thickness of the thinner part joined, so you never end up specifying a fillet larger than the member it is welded to without a specific reason.
• Footnote (c) — cyclic loading. For structures subject to fatigue or cyclic load, the minimum rises to 3/16 in. (5 mm) regardless of thickness. Bridges, crane girders, and moving equipment frames fall here, and it is a common source of shop-floor arguments on thin material.

There is also a special caveat for ASTM A514 and A517 (T-1) quenched-and-tempered steels: these require controlled preheat and interpass temperature per an approved procedure, precisely because their hardenable chemistry magnifies the cracking risk that the minimum-size rules are trying to prevent.

Procedures tie all of this together. If your WPS calls for a low-hydrogen process and qualified preheat, the thinner-part rule in footnote (a) applies and you have room to work; electrode choice alone is not enough. That is why Alloy Steel Welding Electrodes: Selection, Use & Safety treats hydrogen control and consumable storage as sizing issues, not just quality issues.

The practical habit is simple: before striking an arc, identify the thicker part, identify your process and its hydrogen classification, check footnote (a) and footnote (c), and confirm the drawing symbol clears the resulting floor. Per the AISC fillet weld FAQ, if the required minimum conflicts with what the design calls for, the engineer resolves it in writing, not the welder at the bench.

Maximum Weld Size: Why Bigger Is Not Safer

Welders are trained to think of weld size as a minimum problem. The code tells you how small a weld may be, and anything past that feels like free insurance. But the standards also set a ceiling, and ignoring it is one of the more common reasons a good-looking weld turns into a repair order. Oversized welds waste filler metal, amplify distortion, raise residual stress and, along plate edges, can destroy the very throat the design depends on.

The Edge Rule: Thickness Minus 1/16 In

Both AISC 360 Section J2.2b and AWS D1.1 limit fillet weld size along edges of material:

• Material under 1/4 in (6 mm) thick: the maximum fillet size equals the full thickness of the edge.
• Material 1/4 in (6 mm) or thicker: the maximum fillet size equals the thickness minus 1/16 in (2 mm).

There is one escape hatch. If the drawing specifically requires the weld built out to develop the full throat, the welder may exceed that maximum. That decision belongs to the engineer, not the operator on the shop floor.

Why the 1/16 In Rule Exists

The reasoning in the AISC commentary is geometric, not metallurgical. Picture a fillet laid along the edge of a lap plate. Try to make the leg the full thickness of that plate and the arc has nowhere to sit on the upper corner, so it simply melts it away. The weld face still looks correct and a quick visual check may even pass, but the real throat is thinner than the leg size implies. An apparently full-size weld can fall short of the required throat. Holding the leg back by 1/16 in leaves enough edge intact that the inscribed right triangle actually matches the size you think you deposited.

Worked Examples

Plate Edge Thickness	Maximum Fillet Leg (Edge Welds)	Throat at Maximum (×0.707)
3/8 in	3/8 − 1/16 = 5/16 in	≈ 0.177 in
1/2 in	1/2 − 1/16 = 7/16 in	≈ 0.247 in
3/4 in	3/4 − 1/16 = 11/16 in	≈ 0.493 in
3/16 in (under 1/4 in)	Full thickness = 3/16 in	≈ 0.106 in

The pattern is clear. On thin sheet the rule rarely bites, which is why sheet-metal work seldom runs into it. On 3/4 in plate it removes a full sixteenth of leg, and if your drawing called for a 3/4 in edge fillet, the correct response is to build to 11/16 in or request a designed full-throat detail from the engineer.

How Groove Welds Are Limited Differently

Groove welds do not follow the edge rule at all, but they have their own boundaries. AWS D1.1 requires the effective length of a groove weld not to exceed the width of the part being joined, measured perpendicular to the direction of applied stress. The loaded capacity of a groove weld comes from its cross-section, and only the portion inside the joined part contributes. Weld metal spilling past the part edge adds cost and profile problems, not strength.

Reinforcement is capped too. AWS D1.1 limits groove weld reinforcement to roughly 3/32 in (2 mm) on the top and bottom surfaces, and visual quality standards treat reinforcement beyond that as a rejectable condition. Reinforcement above the surface creates a stress concentration at the weld toe, and on fatigue-loaded members it is one of the first things an inspector flags.

When Building Out the Throat Is the Right Move

Sometimes deliberately exceeding the edge maximum is correct. Typical situations include:

• Cyclically loaded joints where the engineer has specified a larger effective throat than the base metal edge can support with a standard fillet.
• Skewed or unequal-leg details where the short leg sits on the thin plate and the long leg must be built up to reach the required throat.
• Repair work where the original throat was compromised and the drawing explicitly authorizes an oversized buildup.

In each case, the enlarged size must appear on the drawing or in a written engineering directive. If the symbol shows a plain fillet along an edge, the maximum governs and the inspector will hold you to it. Sizing decisions, from groove reinforcement limits to the laser-process joints described in alpha laser welding systems, ultimately trace back to what the engineer documented.

The practical takeaway is simple: minimums are metallurgical floors, maximums are geometric ceilings, and a weld that ignores either one is not a stronger weld. It is a rejected one.

Sizing Welds for Load: From Force to Required Throat and Length

Minimums keep a weld from cracking in the puddle; maximums keep it from burning away a plate corner. Neither one tells you whether the weld actually holds the load. That comes from a short, repeatable design path: force in, required throat out, leg size out, length out. Once you have walked it a few times, sizing a fillet becomes almost mechanical.

The Four-Step Design Path

1. Find the load. Get the force the joint must carry, in kips or newtons, from the structural analysis.
2. Convert load to required throat. Fillet weld stress is checked on the effective throat, and that stress is always treated as shear, regardless of the direction of the applied load. Tension, bending, torsion, it all goes through the throat in shear.
3. Convert throat to leg. For an equal-leg fillet, divide the required throat by 0.707.
4. Convert to length. Divide the load by the weld’s capacity per inch of length, then check the minimum-length rules below.

Allowable stress differs by code. AWS D1.1’s basic allowable stress on the effective throat is 0.60 times the nominal tensile strength of the filler metal, which for E70XX filler works out to 0.60 × 70 ksi = 21 ksi. AISC LRFD instead works in load and resistance factors, using available strength tables with a resistance factor of φ = 0.75 for welds.

The Length Rules That Trip People Up

Getting the leg right is only half the job. Three length rules decide whether your weld counts at full value:

• Four-times-size minimum. AWS D1.1 requires that the minimum effective length of a fillet weld be at least four times the nominal size, or the effective size shall not exceed 25% of its effective length. A 1/4 in weld shorter than 1 in gets credited as roughly a 1/16 in weld. Short stitch welds are worse than useless, they are a silent reduction.
• Long-weld reduction. Very long weld groups are less efficient because load does not distribute evenly along them. AISC applies a beta factor when the length exceeds 100 times the leg size: β = 1.2 − 0.002 L/w, not greater than 1.0. A 12 in long 1/4 in fillet (L/w = 48) is untouched; a 60 in long 1/4 in fillet drops to β = 0.95.
• Longitudinal-only joints. Flat bars in tension joined by fillet welds at the ends only must have the transverse weld sized to carry the full force, or the longitudinal welds alone designed for the load, per the AISC treatment of such members.

Worked Example: 20 Kips on a 1/4 In Fillet

Suppose a gusset plate carries a 20 kip load into a flat bar with a 1/4 in equal-leg fillet welded on each side of a lap joint. Using E70 filler under ASD:

• Capacity per inch per weld = 0.707 × 0.25 in × 21 ksi ≈ 3.71 kips/in.
• Two welds = 7.42 kips per inch of combined length.
• Required length = 20 ÷ 7.42 ≈ 2.7 in total, so about 1.35 in per side.
• Check the four-times rule: 4 × 0.25 = 1 in minimum per weld. 1.35 in passes.

Round up to 1.5 in per side for detailing comfort, and the joint works. The same arithmetic in LRFD would use φ = 0.75 on nominal strength instead of the 0.60 allowable, giving a larger nominal capacity but a reduced resistance factor, with a similar final length. The full procedure is documented in AWS’s Fillet Weld Design Handbook and the current AWS D1.1 code.

One practical reminder: adding a second weld line does not double strength if it introduces a long-weld reduction or an eccentric load path. And on aluminum assemblies, where allowable stresses and filler strengths are much lower than steel, the same path applies but the numbers change, as covered in Aluminum MIG Welding: The Complete Practical Guide.

Measuring Weld Size in the Field: Gauges, Techniques, and Acceptance

A confident eye and a tape measure will not get a weld accepted. Field sizing is a gauge job, and the standards treat it that way: under the visual examination framework of AWS B1.11, gauges are explicitly treated as adjuncts to the main visual method, and the acceptance criteria are meant to be defined in writing before the first arc is struck. The welder’s visual inspection handbook used across shipyards follows the same logic: look first, gauge to confirm, record what you found.

The Three Gauges That Cover Almost Every Job

You do not need a drawer full of tools, but you do need the right three:

• Fillet weld gauges. The classic folding set, usually eight leaves, each marked on both faces — one profiled for convex welds, one for concave. Slide a leaf into the joint until it contacts both members; the stamped size it reads is your leg (or your throat, on the concave face).
• Finger gauges. Thin, bent leaves that reach into skewed T-joints, tight lap joints, and out-of-position welds where a flat gauge bottoms out before it reads true. This is also where bead profile knowledge helps: stringer beads and travel-speed-driven profiles read very differently once the joint is skewed, as any detailed Aluminum MIG Welding: The Complete Practical Guide discussion of profile makes clear.
• Bridge cam gauges. The multipurpose instrument. A bridge cam measures fillet leg and throat, groove weld reinforcement, root opening, and — critically — undercut depth, which no fillet set can tell you. If you carry one tool to a fit-up and out-of-position inspection, this is it.

Whatever you use, clean first. Spatter, slag, and on aluminum the ever-present oxide film all fake a bigger weld than actually exists, which makes proper surface preparation as much a measurement step as a welding one.

Technique: Clean, Seat, Find the Smallest Spot

Inspection handbooks and shop-compiled weld quality standards teach a short, repeatable sequence:

1. Clean the weld. Wire-brush or carefully grind away spatter and slag without altering the profile you are about to measure.
2. Seat the gauge flush against both members so both contact points land on base metal, not on the crown of the bead.
3. Gauge where the weld looks smallest. The common field rule is to measure at the location where size visually appears smallest, not along the prettiest stretch of the run.
4. Set a frequency. Welds 5 ft or shorter get gauged once; welds longer than that get gauged at least once every 5 ft.
5. Position your eyes. Stay within about 24 in of the weld and look at 30 degrees or more off the perpendicular, so foreshortening does not make a short leg look generous.

One rule saves beginners every time: on a convex face, gauge the leg; on a concave face, gauge the throat. A convex bead carries extra metal outside the largest inscribed right triangle, so the leg tells the truth about size and the crown is free reinforcement that earns no strength credit. A concave face is the opposite problem — the throat has been eaten away — so measure the shortest throat distance directly, because that is the dimension carrying load.

How Much Undersize the Code Actually Allows

Here is where field judgment turns into a written number. AWS D1.1/D1.1M Table 8.1 permits a fillet weld to underrun its nominal size within limits:

Nominal fillet size	Permitted underrun
3/16 in	1/16 in
1/4 in	3/32 in
5/16 in and larger	1/8 in

Two conditions ride along with those numbers. The undersize portion may not exceed 10% of the weld length in any 12 in of total length, and AWS does not allow a permitted underrun to exist in a stretch shorter than roughly 1 in — a half-inch blip of undersize is a reject, not a tolerance. Oversized welds, by contrast, generally need no acceptance correction as long as they do not interfere with the end use, though the edge-overflow maximums covered earlier in this article still apply.

Done right, the whole check takes under a minute per weld, and it is the difference between a girder that passes the first inspection and a girder that comes back.

Common Weld Size Mistakes and How to Fix Them

Most weld size rejections trace back to a translation problem: something unambiguous on paper got fuzzy between the detailer’s desk, the procedure, and the inspector’s gauge. These are the mistakes that keep recurring in fabricating shops and structural sites, each with the symptom that betrays it and the correction that gets the work accepted.

Reading the Table but Ignoring the Footnotes

AISC 360 Table J2.4 and AWS D1.1 Table 7.7 look like a simple four-row lookup, but footnote (a) in AWS D1.1 decides which thickness you even use, and it turns on process. With non-low-hydrogen electrodes and no qualified preheat procedure, the thicker part governs and the weld runs in a single pass. With low-hydrogen processes, the thinner part governs. Shops that memorize the rows and skip the footnote keep placing welds that are legal for one process and illegal for the other.

Correction: Write the governing process on the traveler beside the required size, and qualify a WPS using the electrode actually in the holder. Consumable selection is where this gets decided, and the fundamentals are covered in Alloy Steel Welding Electrodes: Selection, Use & Safety.

Treating the Code Minimum as the Design Size

The table minimum is a metallurgical floor, not a structural answer. A 5/16 in weld satisfies the thickest row of the minimum table, but if the engineer’s calculation called for 1/2 in, the table is irrelevant and the drawing governs. Welders who “weld to code” instead of “weld to drawing” build a joint that passes a thickness check and fails a load check.

Correction: Verify the symbol size first and the table minimum second. If a drawing calls a size below the table minimum for that thickness, stop and ask the engineer rather than silently upgrading or downgrading.

Running Full Thickness Along a Plate Edge

AISC Section J2.2b caps welds along material edges: full thickness when the edge is under 1/4 in, and thickness minus 1/16 in when the edge is 1/4 in or thicker. Running a full-thickness leg on a 1/2 in plate edge looks generous but is a defect, because the arc melts the upper corner away and the effective throat falls short of what the leg implies.

Correction: Back the leg off to t − 1/16 in, or have the drawing specifically call for the weld to be built out to a full throat. AISC’s welding FAQ explains the reasoning.

Forgetting the 3/16 in Cyclic Floor

Footnote (c) sets a 3/16 in minimum for members under cyclic loading no matter how thin the material is. A 1/8 in weld on a light bracket is fine in static service and wrong on a fatigue-loaded connection. The failure shows up as cracking at the weld toe long after acceptance.

Ignoring Root Openings

A root opening up to 1/16 in passes without modification. Between 1/16 in and 3/16 in, the weld is acceptable only if the size is increased by the root opening or the required effective throat is otherwise demonstrated. A 1/4 in leg across a 3/16 in gap is not a 1/4 in weld, and no amount of convexity recovers the missing throat. Fit-up discipline matters here, especially on aluminum work where gap tolerance is tighter, as explained in Aluminum MIG Welding: The Complete Practical Guide.

Letting Convexity Hide an Under-Throat

A bulging face reads as “plenty of metal,” but strength lives in the throat. Convexity earns no credit, while concavity and undercut subtract directly from it. The gauge rule is simple: measure the leg on a convex face, measure the throat on a concave face.

Short Returns, Short Intermittents, and Skewed Joints

End returns shorter than four times the nominal size get their effective size capped at 25%, and web-to-flange welds on girders may not be undersized for a length equal to twice the flange width at each end. Unequal-leg and skewed joints cause similar trouble: a leg measured on one face says nothing about the throat in a 60-degree T unless the required geometry was sketched.

Correction: Sketch skewed joints, call unequal legs explicitly, and box end returns where the connection depends on them.

Troubleshooting Table: From Symptom to Correction

Mistake	Symptom	Correction
Footnote (a) ignored	Weld sized for the wrong process	Qualify a WPS; note governing thickness rule
Minimum mistaken for design size	Correct thickness, inadequate capacity	Re-read drawing; refer calculation to engineer
Full-thickness edge weld	Melted corner, short throat	Reduce leg to t − 1/16 in or specify built-out throat
Cyclic floor missed	Toe cracking in service	Weld to 3/16 in minimum
Root opening ignored	Leg checks out, throat does not	Increase size by gap or demonstrate required throat
Convexity masking	Big face, small throat	Re-gauge throat; add weld metal over deficient area
Short returns or intermittents	Effective size capped at 25%	Extend welds; box ends per drawing
Skewed or unequal legs assumed	Throat under-design at non-right angles	Revise drawing with a governing sketch

The common thread in every fix is documentation: re-gauge, add weld metal over the deficient area, revise the drawing, or qualify the procedure. Guessing in any of those four directions is how a small size error becomes a rejected assembly.

Choosing the Right Weld Size: Economy, Distortion, and Process Trade-offs

After the load calculations are done and the code minimums checked, a quieter question decides whether the job makes money: how much weld is actually enough? The answer is rarely “add a little extra for safety.” The right weld size is the smallest size that satisfies the structural requirement, clears the code floor, and can be deposited soundly under your procedure.

The Square Law: Why Extra Size Costs More Than It Looks

Fillet weld cross-sectional area grows with the square of the leg length, so cost, deposition time, and distortion rise steeply rather than linearly. The AISC makes this concrete in its engineering FAQs: a 1/2 in fillet weld uses four times as much weld metal as a 1/4 in fillet weld of the same length (AISC Engineering FAQ 8.1.2). Heat input, travel time, filler consumption, and angular distortion all follow that same curve. This is why AISC’s guidance is blunt: the most economical detail uses the least possible fillet weld size consistent with the load.

Practically, that means resisting the habit of rounding every specified 1/4 in weld up to 3/8 in “because it looks better.” A beefier bead looks reassuring on the shop floor, but on a long girder it pulls the flange more, opens gaps between fit-up tack welds, and adds passes nobody paid for.

Fillet or Groove? The Three-Quarter-Inch Crossover

Process economics set a rough crossover point. According to AISC Engineering FAQ 8.1.3, fillet welds smaller than about 3/4 in are generally more economical than equivalent groove welds, because a groove joint requires edge preparation, beveling, and often back-gouging. Above that size, groove welds usually win: a single submerged-arc pass fills a prepared joint faster than a fillet can be built up in layers, and the deposited metal lands directly where strength is needed.

Added penetration can shrink the required leg, but only if it is real. AWS D1.1 historically allowed an effective throat increase for submerged arc welding based on demonstrated penetration, and AWS D1.1 still conditions such credit on qualification. AISC’s position is stricter: any penetration benefit must be demonstrated by testing rather than assumed from a table or a process name. If your WPS does not qualify the extra throat, size the weld on geometry alone.

Distortion, Pass Planning, and Process Choice

Angular distortion follows the heat you put in, and heat follows weld volume. Larger sizes mean more passes: a 3/8 in fillet takes roughly three passes, while 1/2 in and larger grow quickly from there. Each pass reheats the joint and adds another increment of pull, so pass count matters as much as total size. Balancing welds on opposite sides of the joint, using back-step or skip sequencing, and staggering fillets across an assembly all reduce net movement.

Thin or heat-sensitive material pays a worse penalty. Aluminum moves readily under heat, which is why preparation and travel discipline in Aluminum MIG Welding: The Complete Practical Guide matter as much as the bead itself. Multi-pass, out-of-position work also constrains which consumables you can use, and All Position Welding Rod Guide: Types, Uses & Tips covers how rod choice affects how fast you can build size without overheating.

Two Design Mindsets: Strength-Based vs. Optimized Sizing

The simple approach sizes the weld for the calculated load, then stops at the smallest code-compliant size. The optimized approach, set out in the AWS Design Handbook for Calculating Fillet Weld Sizes, also accounts for assembly costs, fit-up tolerance, and the relative economy of welding one member versus the other. That handbook warns in both directions: oversized fillets raise cost and distortion, while undersized fillets drive repair work — and a repair costs far more than the extra ounce of weld metal you thought you were saving.

Issue	What Goes Wrong	Better Move
Over-sized weld	Higher filler cost, more passes, greater distortion, longer out-of-position work	Return to the smallest size that meets load and minimum tables
Under-sized weld	Rejected by inspection, additive repair welds, embedded repair documentation	Re-gauge, add weld metal over the deficient length, revise the drawing if wrong
Unqualified penetration credit	Actual throat below design assumption	Test and qualify the penetration, or size by geometry
Excessive convexity	Reduced effective throat, centerline crack risk at high w/d	Control heat and pass layout for a flatter profile

The practical rule is simple: size welds with the same discipline you size beams. Confirm the load, check the table minimum, respect the edge maximum, and stop there unless the engineer asks for more. When a drawing is ambiguous, or a detail would push you past normal limits, that is the moment to consult the engineer and the current code edition rather than deciding at the arc.

Weld Size Quick-Reference Guide and Conclusion

Weld size has followed us from drawing office to gauge. Here is the article on one page, ready for the tool cabinet.

The One-Glance Reference Table

Reference item	Rule
Minimum fillet size by thicker part	1/8 in up to 1/4 in; 3/16 in over 1/4 to 1/2 in; 1/4 in over 1/2 to 3/4 in; 5/16 in over 3/4 in
Maximum size along a plate edge	Full thickness under 1/4 in; thickness minus 1/16 in at 1/4 in and thicker (unless built out per drawing)
Throat multiplier, equal-leg fillet	Throat = 0.707 × leg; leg = throat ÷ 0.707
Minimum effective length	Four times nominal size, or effective size capped at 25% of length
Undersize allowance	1/16 in on 3/16 in welds, 3/32 in on 1/4 in, 1/8 in on 5/16 in and larger, only where the undersize portion stays under 10% of total length
Gauge frequency	5 ft or shorter: gauge once. Longer: once every 5 ft, at the smallest-appearing spot, eyes within 24 in at 30 degrees or more

The governing documents are AISC 360 Section J2 and its Table J2.4 for design, and AWS D1.1/D1.1M:2025 for fabrication and inspection. In the 25th edition, minimum fillet weld size moved to Clause 7.13 and Table 7.7 from the older Clause 7 spot, and that edition was approved by ANSI in March 2025. It also starts a five-year revision cycle aligned with AISC 360 and the International Building Code, so the edition on your shop wall ages faster than it used to.

The Seven-Step Weld Size Workflow

1. Read the symbol. Size left of the reference line, length and pitch right of it, arrow side versus other side, plus field, box or end-return marks.
2. Check the table minimum against the thicker joined part, then read the footnotes for your electrode process.
3. Check the edge maximum where the weld runs along a material edge, applying the 1/16 in reduction at 1/4 in and above.
4. Convert if the drawing gives a throat: divide required throat by 0.707 for equal legs, use ratio math for unequal.
5. Deposit per the qualified WPS, keeping pass count, heat input and profile inside your qualified procedure.
6. Gauge the finished weld with the right face: leg on a convex profile, throat on a concave one, a finger gauge on skewed or out-of-position joints.
7. Document the reading with location, gauge used and pass/fail status, so the record exists before the member is buried in an assembly.

Three Principles Worth Remembering

• Size is a documented design decision, not a welder’s preference. The engineer sets it, the detailer transfers it, the welder deposits it, the inspector verifies it. Judgment calls belong on the drawing, not in the arc.
• Minimums are metallurgical, maximums are geometric. The floor exists because thick plate quenches a small pool and invites underbead cracking; the ceiling along an edge exists because a welder can melt away the corner and leave an apparently full-size weld short of throat. Strength sizing sits between them.
• The smallest sufficient weld wins. Weld metal volume rises with the square of size, so every extra sixteenth multiplies cost, deposition time and distortion. AISC’s commentary makes the point: the most economical detail uses the least fillet size the code allows.

When the arithmetic is unclear, tolerances conflict, or a symbol contradicts a table, stop and consult the engineer of record rather than improvising. Confirm which code edition governs before you cut a single fit-up, because clause numbers and permitted reductions shift between revisions.

Closing the Loop on That Girder

Remember the 40-foot girder from the start: welds that looked generous to the eye and failed a twelve-dollar gauge over half the run. The legs were short by a sixteenth, exactly the tolerance D1.1 permits in a limited window and exactly the shortfall that kills a design throat calculated at 0.707 of nominal size. The fix was never a heavier weld. It was reading the symbol correctly, checking the table floor, converting throat to leg before the first pass, and gauging every five feet while the weld was still accessible.

That is the discipline of weld size: know the number, know why the code drew the line there, and prove the deposited weld matches. Do that consistently and weld size stops being a source of rework and becomes a specification you meet on the first pass.

Frequently Asked Questions

What does weld size mean on a welding symbol?

For a fillet weld, the size is the leg length of the largest right triangle that fits inside the weld cross-section, written to the left of the fillet symbol. For groove welds, size refers to the required depth of joint penetration, and full-penetration groove welds are usually called out without dimensions.

What is the difference between leg size and throat?

The leg is measured along the members from root to toe, while the throat is the shortest distance from the root to the weld face. On an equal-leg fillet with a flat face, the theoretical throat equals about 0.707 times the leg, and design strength is always calculated on the throat.

What is the minimum fillet weld size allowed by code?

AISC Table J2.4 and AWS D1.1 Table 7.7 set 1/8 in for material up to 1/4 in thick, 3/16 in from over 1/4 to 1/2 in, 1/4 in from over 1/2 to 3/4 in, and 5/16 in over 3/4 in. For cyclically loaded structures AWS D1.1 sets a 3/16 in floor.

Why is there a maximum weld size along a plate edge?

On material 1/4 in thick or more, a fillet along the edge is limited to thickness minus 1/16 in so the corner is not melted away — an apparently full-size weld can otherwise lack the required throat. Under 1/4 in thick, the weld may equal the full thickness.

How do you convert weld size to throat?

Multiply the leg by 0.707 for an equal-leg fillet: a 1/4 in fillet gives a 0.177 in throat and a 1/2 in fillet gives 0.354 in. For unequal legs, multiply the legs and divide by the hypotenuse of the triangle they form.

How much undersize weld is acceptable before repair?

AWS D1.1 Table 8.1 allows underruns of 1/16 in on 3/16 in welds, 3/32 in on 1/4 in welds, and 1/8 in on 5/16 in and larger welds, but only if the undersize portion stays under 10% of the weld length. Beyond that, weld metal must be added to the deficient area.

How do inspectors measure weld size?

They use fillet weld gauges for legs, finger gauges for skewed and out-of-position joints, and bridge cam gauges for throat, reinforcement, and undercut. Gauge the weld where it looks smallest — at least once per weld up to 5 ft, and once every 5 ft on longer welds.

Does a bigger weld always make a stronger joint?

No. Extra convexity adds metal and cost but earns no strength credit because design uses the theoretical throat, while excessive size increases heat input, distortion, and the risk of burning through edges. The smallest weld that meets both the code minimum and the calculated load is preferred.

Should weld size be based on the thicker or thinner member?

It depends on the process. AWS D1.1 Table 7.7 footnote (a) uses the thicker part for non-low-hydrogen electrodes without a qualified preheat procedure, where single-pass welds are also required; low-hydrogen processes use the thinner part.