Skip to main content

A wood screw only works if the wood around it survives the installation. Drill nothing and dense wood splits; drill too big and the threads spin in a smooth hole holding nearly nothing. The right sizes are not folklore, they are published: the screw dimensions come from the ANSI standard reproduced in the NDS, and the percentages come from the same fastener research the code values are built on. Every number below is cited so you can check it.

Read this first. The capacities on this page are reference design values from a published standard, reproduced for education and relative comparison. They already contain the standard's safety margins, but they still assume dry, sound wood, side-grain loading, and correct installation, and they must be adjusted for your conditions before they mean anything about a real connection. For structural work, follow your local code and the fastener manufacturer's published data. When they disagree, the code wins. Modern coated deck and structural screws are a different product with their own tested values; for those, the manufacturer's table is the law.

One screw, three holes

A properly drilled screw joint is three holes, not one. A clearance hole through the piece being attached lets the shank pass without gripping, so the screw can pull the two pieces tight instead of holding them apart. A pilot hole (the NDS calls it a lead hole) in the main member gives the threads wood they can cut into without splitting it. A countersink seats a flat head flush instead of letting it crush its own crater. Skip the first and the threads bridge the joint. Skip the second and dense wood cracks at the worst possible moment, which is always the last screw. Skip the third and the head either stands proud or splits the surface fibres.

Figure 1
Anatomy of a correctly drilled screw joint
tapered tip = 2D no capacity credit countersink 82° cone, head dia. DH clearance hole shank passes free, = D pilot hole 70% to 90% of root Dr, at least as deep as the threads head DH shank D root Dr side grain
Screw dimensions D, Dr, and DH are published in NDS 2018 Appendix Table L3 (ANSI/ASME B18.6.1). The pilot percentages are NDS 12.1.5.2; the tip allowance is Table L3 note 6.

The drill schedule, sized for your screw and your wood

Pick a gauge and a wood. The tool reads the screw's published dimensions from NDS Table L3, sorts your wood into its NDS density band by assigned specific gravity, and returns every hole for the joint as the nearest fractional bit that does not defeat the purpose: pilot bits round down to the next 1/64" so you never give away grip, the clearance bit rounds up so the shank never binds.

Withdrawal and lateral sizes are different, and that is not a typo. For withdrawal (the load tries to pull the screw straight out) the NDS sizes the pilot at 90% of the root diameter in dense wood and 70% in lighter wood, per 12.1.5.2. For lateral loading (the load pushes across the screw) it allows a pilot at the full root diameter in dense wood, per 12.1.5.3, because bearing on the shank does the work and a tighter hole mostly raises the splitting risk. If a joint sees both, drill the withdrawal sizes: they are never larger.

Inputs

Drill schedule

The printable chart

The same numbers for every standard gauge at once. The two pilot columns are the NDS density bands: use the G > 0.6 column for oak, hard maple, hickory, beech, and other dense hardwoods, and the G ≤ 0.6 column for construction softwoods, poplar, and cherry-density stock. In wood with G ≤ 0.5 the NDS does not require a withdrawal lead hole at all (12.1.5.2); drill it anyway within a couple of inches of an end or an edge, because a split there costs more than the ten seconds of drilling.

Wood screw pilot, clearance, and countersink sizes by gauge (NDS 2018 Table L3 dimensions)
GaugeShank DRoot Dr Pilot, G ≤ 0.6
softwoods, 70% Dr
Pilot, G > 0.6
dense hardwoods, 90% Dr
Clearance
side member
Countersink
82° head
#60.138"0.113" 5/64" (2.0 mm) 3/32" (2.4 mm) 9/64" (3.6 mm) 17/64"
#70.151"0.122" 5/64" (2.0 mm) 7/64" (2.8 mm) 5/32" (4.0 mm) 19/64"
#80.164"0.131" 5/64" (2.0 mm) 7/64" (2.8 mm) 11/64" (4.4 mm) 5/16"
#90.177"0.142" 3/32" (2.4 mm) 1/8" (3.2 mm) 3/16" (4.8 mm) 11/32"
#100.190"0.152" 3/32" (2.4 mm) 1/8" (3.2 mm) 13/64" (5.2 mm) 3/8"
#120.216"0.171" 7/64" (2.8 mm) 9/64" (3.6 mm) 7/32" (5.6 mm) 27/64"
#140.242"0.196" 1/8" (3.2 mm) 11/64" (4.4 mm) 1/4" (6.4 mm) 31/64"
#160.268"0.209" 9/64" (3.6 mm) 3/16" (4.8 mm) 9/32" (7.1 mm) 33/64"
#180.294"0.232" 5/32" (4.0 mm) 13/64" (5.2 mm) 19/64" (7.5 mm) 39/64"
#200.320"0.255" 11/64" (4.4 mm) 7/32" (5.6 mm) 21/64" (8.3 mm) 5/8"
#240.372"0.298" 13/64" (5.2 mm) 17/64" (6.7 mm) 3/8" (9.5 mm) 47/64"

Pilot bits are the computed percentage of root diameter rounded down to the nearest 1/64"; clearance and countersink round up. Shank, root, and head diameters from NDS 2018 Appendix Table L3. Screws smaller than #6 have no published root diameter in Table L3; measure the root with calipers and apply the same 70%/90% rule. At 3/8" diameter and up you are in lag screw territory, which has its own hole rules: see the structural lags & bolts guide.

How much of your screw is actually holding?

Length is the number on the box, but capacity comes from engaged thread: the threaded length that ends up inside the main member, minus the tapered tip, which the standard lets you assume is two diameters long and carries nothing. A 2" screw driven through 3/4" stock does not give you 2" of holding, it gives you about 0.9" of working thread. The NDS turns that into a design value with one compact formula:

W = 2850 × G² × D NDS Eq. 12.2-2 — reference withdrawal design value, lb per inch of thread penetration into side grain

Two things are worth internalising. Capacity scales with G squared, so wood density matters twice over: the same #10 screw is worth 95 lb per inch of thread in spruce-pine-fir and 288 lb per inch in white oak, three times as much, from the density term alone. And the diameter term is linear, so going up two gauge numbers buys roughly 15% more capacity while a denser wood species can buy you 100%. If a withdrawal connection is marginal, the wood is usually the lever, not the screw.

The calculator below does the whole chain: thread length from the screw standard, penetration from your geometry, tip deducted, minimum penetration checked against the NDS floor of six diameters (12.1.5.6), and the result adjusted for load duration and wet service. It also runs the check almost everyone skips: whether your screw's unthreaded shank actually spans the piece you are clamping.

Inputs

Results

Side-grain withdrawal only. The NDS assigns wood screws loaded in withdrawal from end grain a factor of zero (12.2.2.3): no design capacity, no exceptions. Lateral (shear) capacity is a separate calculation with its own tables (NDS 12.3 and Tables 12L–M), and on very hard screws driven deep, the screw's own tensile strength at the root can govern before the wood lets go (12.2.2.5).

The mistake the clearance hole prevents

Threads are a one-way ratchet for wood. If the threaded part of the screw is engaged in both pieces, the screw cannot pull the joint closed: the threads hold each board exactly where they first bit, gap included. Tightening harder just strips the top piece or snaps the screw. This is why a screw that is "long enough" can still be the wrong screw: if the unthreaded shank is shorter than the piece you are attaching, the threads cross the joint line.

Figure 2
Why threads must not cross the joint line
1 Threads bridge the joint
The gap is locked open
gap locked threads grip here... ...and here
2 Clearance hole drilled
The head clamps the joint shut
tight shank spins free threads grip only here
The engagement calculator flags this automatically: if the unthreaded shank is shorter than the attached piece, the threads cross the joint and the clearance hole stops being optional.
Worked example — #8 × 2" screw through 3/4" stock into Douglas Fir-Larch
1 Look up the screw. A #8 has shank D = 0.164" and root Dr = 0.131" (NDS Table L3). Douglas Fir-Larch has assigned G = 0.50 (Table 12.3.3A), which lands in the G ≤ 0.6 band: pilot at 70% of root, 5/64" bit, 11/64" clearance through the 3/4" piece.
2 Find the working thread. A cut-thread screw is threaded for 2/3 of its length: 1-1/3". Penetration past the joint is 2" − 3/4" = 1-1/4", which clears the NDS minimum of 6D = 0.98" (12.1.5.6). Deduct the 2D tip, 0.33", and the engaged thread is 0.92".
3 Read the unit value. W = 2850 × 0.50² × 0.164 = 117 lb per inch. Table 12.2B agrees: it tabulates 117 at G = 0.50 for a #8.
4 Multiply. 117 × 0.92 = 108 lb allowable withdrawal, dry service, normal duration.
About 108 lb for that one screw, and note what the same math says about the shorter screw you almost grabbed: a #8 × 1-1/2" through the same 3/4" stock penetrates only 3/4", under the 6D minimum of 0.98". It fails the standard before it holds a single pound. The jump from 1-1/2" to 2" is the difference between a connection the NDS recognises and one it does not.
These are allowable values, not breaking strength. A reference design value already carries the standard's safety margin, so it sits well below the load where the screw actually tears out in a test. Do not stack your own factor on top, and do not read it as the failure load. For sustained loads, remember the other direction too: dead load runs at CD = 0.9, so a screw holding something permanently holds 10% less on paper than the normal-duration number.

Rules that surprise people

End grain gets zero. Not less, zero.

The NDS assigns withdrawal from end grain Ceg = 0.0 (12.2.2.3): no design value, no exceptions. Lab tests average around 75% of side-grain values when nothing splits, which is exactly the problem, you cannot count on nothing splitting. If a load must hang from end grain, change the joint: dowel across it, threaded insert, or a mechanical fastener into cross grain.

Soap on the threads is engineering, not cheating.

NDS 12.1.5.5: lubricating the screw or the lead hole causes no reduction in design value, and the Wood Handbook recommends it outright for dense woods. The friction you remove is friction that was fighting insertion, not friction that was holding your joint.

A screw is turned, never driven.

Hammering a wood screw home wrecks the thread-to-fibre interface the withdrawal values assume (NDS 12.1.5.4). If the screw needs that much persuasion, the pilot hole is too small or too shallow. Drill it right and drive it with a driver.

Wet wood costs you 30%.

Withdrawal in wood that stays above 19% moisture runs at CM = 0.7 (Table 11.3.3). Fence rails, planters, anything ground-adjacent: size the connection for the wet number, not the showroom number, and use fasteners rated for the treatment chemistry.

Where this page hands off

Everything above is the wood screw story. At 3/8" shank diameter you are in lag screw territory, where the hole rules change shape: a stepped lead hole at 40–85% of the shank by density band, a 4D penetration floor, and a wrench instead of a driver. For the argument about when a screw is the wrong fastener entirely, the nails vs. screws page covers the brittle-failure trade. Dimensions and design values for the whole fastener family live in technical specifications, and the assumptions behind every number we publish are laid out in the engineering methodology.

Frequently asked questions

What size pilot hole for a #8 wood screw?

7/64" in dense hardwood (oak, hard maple, hickory) and 5/64" in softwood, with an 11/64" clearance hole through the piece being attached. Those come from the #8's published root diameter of 0.131": the NDS sizes withdrawal pilots at 90% of root in wood with G above 0.6 and 70% below it.

Do I always need a pilot hole?

In dense hardwood, always. In softwoods with G at or below 0.5 the NDS does not require a lead hole for withdrawal loading (12.1.5.2), and production framing mostly skips them. Drill one anyway within a couple of inches of an end or edge, in anything that splits like cedar, and for every screw whose placement you cannot afford to lose. The clearance hole through the attached piece is a separate question, and the answer there is yes whenever the threads would otherwise cross the joint line.

Why does this chart disagree with the chart on my drill bit case?

Most retail charts print one pilot size per gauge, usually close to the full root diameter, which is the NDS lateral-load sizing for dense wood. This chart separates the cases the standard separates: withdrawal pilots at 70% or 90% of root by density band, lateral pilots at the root itself. A one-size chart is not wrong for hanging shear loads in hardwood; it is giving away withdrawal grip everywhere else.

How deep should a pilot hole be?

At least as deep as the threads will penetrate, so the full threaded length cuts into pre-relieved wood. Wrap a flag of tape around the bit at the target depth. Too shallow forces the tip to wedge unrelieved fibres apart, which is how "the last quarter turn" splits a rail.

How much weight can a wood screw hold?

In withdrawal, the NDS reference value is 2850 × G² × D pounds per inch of engaged thread. A #10 with 1" of working thread holds a design value of about 95 lb in spruce-pine-fir and about 288 lb in white oak, before load-duration and moisture adjustments. Lateral (shear) capacity is a different calculation, and for hooks, shelves, and brackets the joint usually sees both at once, so run the actual geometry in the calculator above rather than quoting one number.

Can I put a screw into end grain?

Not for withdrawal loads. The NDS assigns end-grain withdrawal a factor of exactly zero (12.2.2.3), because the threads shear the fibre bundles apart instead of hooking across them. Tests that avoid splitting average about three-quarters of side-grain strength, but design practice refuses to rely on it. Use a dowel or threaded insert to give the screw cross grain to bite.

Does soap or wax on the threads weaken the joint?

No. NDS 12.1.5.5 says no reduction in design value is anticipated from lubricating the screw or the lead hole, and the Wood Handbook recommends it for dense woods. It lowers driving torque, which mostly protects the head recess and your patience.

What is the difference between cut thread and rolled thread screws?

A cut-thread screw is machined from full-diameter stock and threaded for about 2/3 of its length. A rolled-thread screw is formed from smaller stock, threads at least four diameters long or 2/3 of the length, whichever is greater, and its root is close to 80% of the nominal shank. Both use the same NDS withdrawal equation; what changes is the geometry, which is why the calculator asks.

Sources

  • NDS 2018 — AWC National Design Specification for Wood Construction. Appendix Table L3 (wood screw dimensions, thread lengths, 2D tip allowance); 12.1.5 (lead holes, insertion, 6D minimum penetration); Eq. 12.2-2 and Table 12.2B (withdrawal design values); 12.2.2.3 (end grain, Ceg = 0); 12.2.2.5 (root tensile limit); Table 12.3.3A (assigned specific gravities); Table 11.3.3 (wet service CM); Table 2.3.2 and 11.3.2 (load duration CD).
  • Wood Handbook 2021 — USDA Forest Products Laboratory, FPL-GTR-282, Chapter 8 (Fastenings): 70%/90% of root diameter lead-hole practice, ultimate withdrawal test equation, end-grain test behaviour, tapping-screw comparison, lubrication guidance.
  • ANSI/ASME B18.6.1Wood Screws, the dimensional standard NDS Table L3 reproduces; the source of the D = 0.060 + 0.013 × gauge relationship.

The NDS and the Wood Handbook are both in this site's reference library and every value above was extracted from those documents directly, not from secondary charts. Notation: G is assigned specific gravity, D shank diameter, Dr root diameter, W reference withdrawal design value per inch of thread penetration.