Structural Screws vs. Lag Bolts: Can a 5/16 Screw Replace a 1/2-Inch Lag?
Structural Screws vs. Lag Bolts: Can a 5/16 Screw Replace a 1/2-Inch Lag? - A step into the workshop.
Stop Wasting Wood
Generate clear cut diagrams instantly with our new Cutlist Optimizer. Plan your cuts to make your next build easier.
Every August the same conversation happens at a lumber counter somewhere in Texas. Somebody's setting pergola posts, somebody else says skip the lags, a 5/16 structural screw does the same job, drive it and go home. It's a widely circulating shorthand, and it compares shear on one fastener to tension on the other while citing paperwork almost nobody repeating it has opened.
So I opened it. What's published is more interesting than the shorthand, and in one direction a lot worse.
Two Agencies, Not One
Half this argument leans on a report that doesn't exist. GRK's RSS Rugged Structural Screw is covered by ICC Evaluation Service report ESR-2442, reissued October 2025, revised March 2026, up for renewal October 2027, held by GRK Fasteners, a division of Illinois Tool Works. That one's real.
Simpson's SDWS and SDWC aren't ICC-ES at all. They're evaluated by IAPMO Uniform Evaluation Service: SDWS in ER-192, revised February 2026, SDWC in ER-262, revised June 2026. Different accredited body, different numbers. Simpson does hold an ICC-ES report — ESR-3046, reissued August 2026 — but it covers the SD, SDWF, SDCP, SDCF and SDHR families and carries no SDWS or SDWC design tables. Go hunting there for Timber screw numbers and you'll come up empty.
The value entry is ICC-ES: Grip-Rite Structural Screws, report holder PrimeSource, ESR-3933, reissued April 2026. With a trap attached. The ESR-3933 PDF on Grip-Rite's own website is the April 2021 edition, cover marked subject to renewal April 2023 — expired paperwork on the manufacturer's server. Its withdrawal and lateral tables match the current edition, so nobody got burned on those numbers. Doesn't matter. Pull reports from the evaluation body, not the brand's PDF library.
Worth knowing what ESR-2442 actually authorizes as a swap, too. Its only substitution table trades RSS screws for nails — a 5/16 RSS stands in for a 16d common. There's no lag substitution table anywhere in it. GRK's own product page calls the RSS an "easy to install lag alternative" and publishes no comparative figures against any lag. Fine Homebuilding makes the same point from the code side: IRC Table R502.2.2.1 lists half-inch bolts and half-inch lags for ledgers, and proprietary screws aren't in it.
A 5/16 That's Nowhere Near 5/16
ESR-2442 says it out loud: Dnom is "Fastener size designation used by the applicant." Marketing name. Real geometry gets listed separately, and GRK's "5/16-inch" RSS measures 0.276 inch across the outside of the thread, 0.195 at the shank, 0.167 at the root. Nothing on it is 0.3125 anywhere. On a lag the size designation is the shank diameter, and the National Design Specification tables are built on it. The popular comparison sets a measured diameter against a marketing one.
ESR-2442 §4.2.3 is blunt about the consequence: an engineer running the NDS yield equations has to use Dr, the root, everywhere the NDS says D. The design diameter of a 5/16 RSS is 0.167 inch.
It gets funnier across brands. Per ER-192 Table 1, Simpson's SDWS22 has a 0.221-inch shank and a 0.315-inch major thread. GRK's "3/8" RSS runs 0.313 outside the thread. Same thread diameter, names two sizes apart.
What the screw does have is steel. ESR-2442 Table 1 rates the 5/16 RSS at 171,800 psi bending yield, allowable tension 1,274 lbf, allowable shear 884 lbf. The American Wood Council's NDS 2018 Table 12J, footnote 2, says its lag values assume 60,000 psi at 5/16 inch and 45,000 psi at 3/8 and larger. Call it 2.9 times the bending yield. That's the whole trick. Not magic, just harder wire.
Shear: The Shorthand Half Works
Reference lateral values, single-shear wood-to-wood, load duration factor of 1.0, 1-1/2 inch side member:
| Fastener, 1-1/2 in. side member | SPF, G 0.42 | DF-L, G 0.50 |
|---|---|---|
| GRK 5/16 RSS, parallel (ESR-2442 Table 3) | 239 lbf | 333 lbf |
| 1/2 in. lag, parallel (NDS Table 12J) | 350 lbf | 390 lbf |
| Simpson SDWS22600DB (ER-192 Table 5) | — | 405 lbf (DF/SP) |
In spruce-pine-fir the GRK 5/16 delivers 68 percent of the half-inch lag parallel to grain. In Douglas fir-larch, 85 percent. Simpson's SDWS22600DB in Douglas fir or southern pine hits 405 against the lag's 390, so that one genuinely does match. And perpendicular to grain in SPF, the GRK gives 236 lbf against the lag's 170 — it beats the lag by 39 percent.
Honest version: a 5/16 structural screw runs somewhere between 68 and 104 percent of a half-inch lag laterally, depending on brand, species and grain direction, and it wins outright across the grain.
The clean "replaces a half-inch lag" line has a traceable origin. A fastener blog at princefastener.com tabulates 2,948 lb of ultimate shear for a 5/16 GRK RSS and credits ESR-2442. That report has no such number — Table 1 says 884 lbf allowable shear, Table 3 tops out at 472 — and the same page cites "NDS Table 12.2C" for lag withdrawal, which is the nail and spike table. Origin of the myth, not evidence for it.
Withdrawal: Not Close
Withdrawal is tension. The pergola beam trying to lift off the post, the stretcher pulling out of a bench leg. Here the lag wins and it isn't a debate.
Per inch of thread penetration, NDS Table 12.2A gives a half-inch lag 291 lbf in SPF, 378 in Douglas fir-larch, 437 in southern pine. A 5/16 lag gets 205 in SPF. ESR-2442 Table 2 gives the 5/16 RSS 165 lbf per inch at SG 0.42 and 227 at SG 0.55. ER-192 puts the SDWS22600DB at 214 in Douglas fir and southern pine, 57 percent of the lag. Grip-Rite's 5/16 manages 119 at SG 0.42.
In SPF the half-inch lag is 1.76 times the GRK RSS per inch, and a same-size 5/16 lag still beats it by 24 percent. The screw doesn't replace a lag of its own nominal size in tension, let alone one 60 percent bigger. The only thumb on the screw's scale: the NDS counts lag thread excluding the tapered tip, ESR-2442 counts embedded thread including it.
Then the cap nobody brings up. ESR-2442 Table 2 publishes head pull-through separately — a 5/16 RSS through a 1-1/2 inch side member gets 356 lbf at SG 0.42, 504 at 0.50, 610 at 0.55. Run a 5/16 by 4 inch RSS into SPF with roughly 2-1/2 inches of thread buried: 165 times 2.5 is 413 lbf of withdrawal against 356 lbf of pull-through. The head governs. Step up to the 6-inch screw with 3-7/8 inches of thread and withdrawal climbs to 639, while pull-through sits right where it was. A longer screw buys nothing in tension once the head is the weak link. The fix is a thicker side member or a washer.
Simpson bakes this in. ER-192 Table 7 carries a WMAX column defined as the lesser of withdrawal or pull-through of a 1-1/2 inch side member — 590 lbf in DF/SP, 495 in hem-fir and SPF, for the 5-inch model and up. Same physics, far lower odds of missing it.
Wet Changes the Answer Unequally
Every number above is dry-service. On outdoor furniture, a pergola, an open shed in a Texas summer, in-service moisture goes past 19 percent and the wet service factor lands. The reports don't agree on the penalty. GRK applies 0.70 to withdrawal and lateral both, Simpson's ER-192 applies 0.70 to lateral above 19 percent, and Grip-Rite's ESR-3933 applies 0.46 to withdrawal and 0.51 to lateral.
Dry, GRK's 165 lbf per inch beats Grip-Rite's 119 by 39 percent. Wet, it's 115.5 against 54.7 — more than double. A value screw that looks close on a dry spec sheet is half the fastener in a pergola. No report explains the gap, so don't read it as bad steel. Read it as published, and design to it.
Coating scope is narrower than people assume, too. ESR-2442 evaluated Climatek in copper azole to 0.15 pcf retention, ER-192 evaluated Simpson's coatings against ACQ-D at 0.40 pcf, and both stop at freshwater. Section 5.6 of ESR-2442 puts saltwater and salt spray outside the evaluation entirely — coastal furniture is off the edge of these documents.
What a Connection Costs
On furniture-scale parts the first cost isn't money, it's room. ESR-2442 Table 5 wants 15D end distance and 10D edge distance for a self-drilled RSS in wood under SG 0.50, rising to 20D and 12D above it; predrilling cuts those to 12D and 7D. Simpson publishes inches instead — ER-192 Table 16 wants a 6-inch end distance for a laterally loaded SDWS22, which is longer than most bench legs are wide. Capacity is rarely what governs on a small part.
Prices as of 18 August 2026. The Fastener Depot lists a 1/2 by 4 inch hot-dipped galvanized hex lag at $0.98 each, Chinook Lumber lists GRK 5/16 by 4 inch RSS at $73.39 per hundred or $0.73 each, and Confast lists the Simpson SDWS22600DB at $22.24 a dozen, $1.85 each.
Match capacity instead of counting screws. One half-inch lag gives 350 lbf parallel in SPF; getting there with GRK RSS at 239 lbf each takes 1.46 screws, so two. That's $1.47 against $0.98 — fifty percent more material, and it buys back the washer and the two-part pilot hole NDS 12.1.4.2 requires for a lag.
The number nobody prints is cost per unit of wet capacity. That GRK screw at $0.734 delivers 239 times 0.70, or 167 lbf of wet lateral capacity. Forty-four cents per hundred pounds. Outdoors, that's the metric. Grip-Rite I can't price against capacity honestly: the verified listing at Hardware World, $12.55 for 20, is a 3-inch screw, and the 191 lbf row in ESR-3933 needs 2-3/8 inches of main-member penetration it can't reach.
The Call
Screws where the load is lateral and you're driving a lot of connections. Lags, with washers, where the joint is in tension — pergola beams on posts, anything trying to lift. And check the head before you buy a longer screw, because past a point you're paying for thread the head can't hold.
One more thing to sit with. Every screw number here traces to a manufacturer-sponsored evaluation under the same acceptance criteria; the lag numbers come from the American Wood Council's consensus NDS. The reports are the best public record there is, and they still aren't a referee. Design to the published value, not to the counter guy's ratio.