Counterbore and Countersink Calculator

Counterbore diameter and depth for a socket head, countersink top diameter and depth for a flat head, the material left under either, and the sheet-thickness limit that actually binds — which on thin sheet is the minimum land in millimetres, not any percentage. The 82°/90° fork is worked out exactly, both ways round, with each standard’s angle recovered from its own published head heights.

Last updated 24 Sep 2026

Counterbore and countersink

Screw and angle → recess diameter and depth

Screw sizeHead dimensions are ISO 4762 for the counterbore case and ISO 10642 / DIN 7991 for the countersink case.

Which recessA counterbore swallows a cylindrical head; a countersink seats a conical one. The headline is the recess DIAMETER either way — the number that goes on the drawing first.

Countersink included angleThis is a standards fork with a physical consequence: a screw in the wrong-angle recess bears on a LINE instead of a cone. The calculation below says which way and by how much.

Clearance hole class (ISO 273)The hole under the recess. It sets where the cone stops, and it is also the diameter a countersink’s own depth is measured against.

How far below flush you want the head (mm)Zero is flush. A counterbore is usually cut 0.5 to 0.8 mm deeper than the head so the head is certainly below the surface; a countersink cut below flush loses bearing area under the head, which is the opposite problem.

Sheet or part thickness (mm)Used for the material left under a counterbore, and for the sheet-thickness limit on a countersink.

Material, for the countersink depth limitThree published limits, three different numbers. The page prints all of them and tells you which one is actually binding on your thickness — because on thin sheet it is the minimum land in millimetres, not any percentage.

Not a circuit: one clearance hole with BOTH countersink angles drawn over it, and the head that cannot seat drawn where it actually stops. The two V outlines share a top diameter and a hole: the DASHED one is 90°, the metric norm, and the solid one reaching further down is 82°, the inch norm. Scaled so the head's theoretical diameter fills the same width at every size, so what changes when you step the size is the proportions and not the picture. The block sitting above the surface is an 82° head placed where the geometry puts it: its lower edge jams on the 90° wall, because an 82° flank falls away more slowly than a 90° bore narrows, so the head cannot descend and its top face stands proud. Cutting the 90° cone deeper does not help — the head just goes further in without ever touching the flank. On the right the same gap — between the head's top face and the part surface — is drawn again at TEN TIMES the scale, as two blocks with the gap dimensioned between them, because half a millimetre on an M6 is a hairline beside a 13 mm head and a figure you cannot see is a figure nobody checks. The hatched block there is the part and the plain one is the head. The reverse mismatch — a 90° head in an 82° cone — is not drawn because there is nothing to draw: it sits flush and bears on a line at the rim, which is the failure that passes inspection.

14.44mmExample

An M6 flat head in a 90° countersink, cut for 0.5 mm below flush over a medium clearance hole in a 6 mm plate

A cone, a cylinder, and an angle that has to match

counterbore Ø = head dk + clearance; depth = head k + below-flush · countersink Ø = dk + 2·(below-flush)·tan(θ/2) · depth = (Ø − d_hole) / (2·tan(θ/2)) · an 82° head stands proud of a 90° bore by (dk − d_hole)/2 · (cot 41° − 1)

θ: the included angle. 90° is the metric norm and 82° the inch one, and the choice is not a preference — a head and a recess of different angles cannot make conical contact at all
dk: the head’s theoretical maximum diameter. Theoretical because the real head is truncated: ISO 10642 prints an actual minimum head diameter well below dk, and the cone the recess has to match is defined by dk
k: head height. For ISO 10642 it is exactly (dk − d)/2, which is how the 90° is confirmed; for ISO 4762 it is exactly the nominal diameter
d_hole: the clearance hole under the recess. It is where the cone stops, so it is in every countersink formula on this page
the proud amount: derived here and checked by brute force — sample both cone profiles densely and find the deepest the head can descend without the profiles crossing. The closed form and the brute force agree to a ten-thousandth of a millimetre at five geometries

Worked example

An M6 flat head in a 90° countersink, cut for 0.5 mm below flush over a medium clearance hole in a 6 mm plate
The ISO 10642 M6 head's theoretical diameter is 13.44 mm and its published height is 3.72 mm. Check the angle from those two numbers alone: (13.44 − 6)/2 = 3.72, and (dk − d)/2 is the height of a 45° half-angle cone. The 90° is in the dimension table, not just in a drawing note
For 0.5 mm below flush the cone has to open up by 2 × 0.5 × tan 45° = 1.00 mm, so the top diameter is 13.44 + 1.00 = 14.44 mm
Its depth, from the 6.6 mm clearance hole, is (14.44 − 6.6) / (2 tan 45°) = 3.92 mm. At exactly flush it would be 3.42 mm
In 6 mm steel the 80% rule allows 4.80 mm and the 0.25 mm land rule allows 5.75, so the percentage binds and 3.92 mm is inside it. In 1 mm sheet neither would be: the land rule would cap the depth at 0.75 mm and an M6 flat head simply does not fit in it
Now the mismatch. Put an ASME B18.3 82° head of the same theoretical diameter into this 90° cone and it CANNOT descend: an 82° flank falls away more slowly than a 90° bore narrows, so the head's lower edge jams. It stands proud by (13.44 − 6.6)/2 × (cot 41° − 1) = -0.514 mm. Half a millimetre on an M6, and cutting the 90° cone deeper does not fix it — the head just goes further in without ever touching the flank
The other way round is worse because it looks right. A 90° head in an 82° cone of the same top diameter sits flush, and bears on a LINE at the rim of the countersink rather than on the cone at all

The 82°/90° fork, and how the angle is recovered from the published head heights

Columns three and four are the point. ISO 10642’s published head height is EXACTLY (dk − d)/2 at every size from M3 to M20 — and (dk − d)/2 is the height of a cone with a 45° half angle and nothing else. So the 90° included angle is not something you have to take on trust from a drawing note; it is in the dimension table. The same check works the other way for ASME B18.3: its published flat-head heights are exactly (dk − d)/(2 tan 41°) at all thirteen sizes, to the last printed digit, which is an 82° cone. Column five shows what the difference costs. A 90° head is shorter than an 82° head of the same top diameter by that much, which is why the last column exists: seat an 82° head in a 90° bore cut to the same top diameter and it cannot descend — the bore narrows faster than the head does — so it stands proud by the amount shown, half a millimetre on an M6. The reverse is not symmetric and it is worse: a 90° head in an 82° bore of the same top diameter cannot sink at all either, and it bears on a LINE at the rim of the countersink rather than on the cone. That is the worst bearing condition available to a screw. The same designation can mean different dimensions in different standards families — ANSI against ISO, inch against metric, one national standard against another. The family used here is named beside every figure; check which one your part was made to.
Screw	ISO 10642 dk max (mm)	Published k max (mm)	(dk − d)/2 — the 90° cone’s height	Height an 82° cone would have (mm)	Difference (mm)	How far an 82° head stands PROUD of a 90° bore of the same top diameter (mm)	dk ÷ d
M3	6.72	1.86	1.86	2.14	-0.28	0.250	2.240
M4	8.96	2.48	2.48	2.85	-0.37	0.335	2.240
M5	11.20	3.10	3.10	3.57	-0.47	0.429	2.240
M6	13.44	3.72	3.72	4.28	-0.56	0.514	2.240
M8	17.92	4.96	4.96	5.71	-0.75	0.671	2.240
M10	22.40	6.20	6.20	7.13	-0.93	0.857	2.240
M12	26.88	7.44	7.44	8.56	-1.12	1.006	2.240
M16	33.60	8.80	8.80	10.12	-1.32	1.210	2.100
M20	40.32	10.16	10.16	11.69	-1.53	1.377	2.016

Counterbore for an ISO 4762 socket head cap screw

Two clean facts about ISO 4762 fall out of this table and are worth remembering: the head height is EXACTLY the nominal diameter at every size, and the head diameter is exactly 1.5 d from M12 up (it runs a little wider below that, 1.83 d at M3). So a counterbore for a socket head is 1.5 d across and d deep, plus clearance, and you can size one in your head. The published counterbore diameters come from two publishers who agree on every diameter and differ on the depth — 6.5 mm against 6.8 mm at M6 — because the depth is the head height plus an allowance each of them chose. Hence the two depth columns. Whole-Spec’s own note is worth keeping: making the counterbore diameter exactly the head diameter “is a mistake” — plating, a fillet under the head and the tool’s own runout all need somewhere to go. These dimensions come from a published standard’s table, not from a formula. The standard itself is cited below and the printed values are attributed to the catalogue they were taken from; a different publisher may round differently in the last digit.
Screw	Head dk max (mm)	Head k max (mm)	Medium clearance hole (mm)	Published counterbore Ø (mm)	Diametral clearance	Depth for 0.5 mm below flush	Depth for 0.8 mm below flush	dk ÷ d
M3	5.5	3.0	3.4	6.5	1.0	3.5	3.8	1.833
M4	7.0	4.0	4.5	8.0	1.0	4.5	4.8	1.750
M5	8.5	5.0	5.5	9.5	1.0	5.5	5.8	1.700
M6	10.0	6.0	6.6	11.0	1.0	6.5	6.8	1.667
M8	13.0	8.0	9.0	14.0	1.0	8.5	8.8	1.625
M10	16.0	10.0	11.0	17.5	1.5	10.5	10.8	1.600
M12	18.0	12.0	13.5	20.0	2.0	12.5	12.8	1.500
M16	24.0	16.0	17.5	26.0	2.0	16.5	16.8	1.500
M20	30.0	20.0	22.0	33.0	3.0	20.5	20.8	1.500

The two errors, and the window between them

Two of these five are angle mismatches and only one of those is visible. That is the reason the angle is worth printing on the drawing rather than assuming: the 82° head in a 90° hole announces itself by standing proud, and the 90° head in an 82° hole sits down looking perfect while bearing on a circle. This page sizes a part; it does not certify one. Where the answer carries a consequence — a load path, a lifting duty, a pressure boundary, a fastener holding something that can fall — confirm it against the design code that governs the application, and against the manufacturer’s own rating, before relying on it.
Error	What it looks like	What it costs	The rule
Countersink cut TOO SHALLOW	The head stands proud of the surface. Obvious on inspection, and usually caught	The joint never pulls up: the head bottoms on the cone before the underside of the head reaches the surface, so all the preload goes into wedging the head into the cone rather than clamping the parts. On a sheet joint it also means the mating part cannot sit flat	Cut until the head’s top face is at or a few hundredths below the surface. Measure the head, not the nominal, because the theoretical dk is larger than the actual head
Countersink cut TOO DEEP	The head disappears below flush. Looks tidier than the shallow error and is more dangerous	Two separate losses. The bearing area under the head grows on paper but the LAND under the recess shrinks, so in sheet the cone eventually reaches a knife edge and there is nothing left to bear against; and the head can bottom on the clearance hole’s own rim rather than on the cone, which is a line contact	The deeper of the percentage rule and the minimum-land rule, computed above. On thin sheet the land is the binding one
Wrong INCLUDED ANGLE — 82° head in a 90° recess	The head sits proud, and it looks like the shallow error	Half a millimetre on an M6, and it is not fixable by going deeper: cut the 90° cone deeper and the head just descends further without ever seating on the flank, because the flank angles are different. The contact stays a line at the head’s lower edge	Match the angle. 90° for metric (ISO 10642 / DIN 7991), 82° for inch (ASME B18.3, ±2°)
Wrong INCLUDED ANGLE — 90° head in an 82° recess	The head appears to sit flush, which is why this one ships	The head bears on a LINE at the rim of the countersink, not on the cone. All the preload goes through a circle of contact instead of a conical band, so the local pressure is enormous and the rim yields. Cut the 82° recess wider to make the head sink and it drops below flush without ever gaining conical contact	Match the angle. This is the failure that looks correct
Counterbore cut too deep in a thin part	Nothing, until the part is loaded	The material left under the bore is what carries the bolt’s bearing load into the part. Cut a 6 mm deep counterbore in a 8 mm plate and 2 mm is carrying everything	This page computes the material left. Keep it at least half the screw diameter as a first pass, and check it against the bearing stress properly if it is close

How deep a countersink may go in sheet — and which rule actually binds

There is no single 70% rule, and this batch could not find one. What is published is material-dependent and comes with a second criterion. PASS Stanztechnik give, for a 90° countersink, “max. 90% of s” for aluminium, “max. 80% of s” for steel and “max 66% of s” for stainless, each with a minimum cylindrical part or depth — 0.25 mm generally and 0.55 mm for stainless. SendCutSend, a job shop rather than a standards body, publish a single stricter figure: the countersink “should not exceed 60% of material thickness”, with a 0.125 in minimum sheet thickness and a target of “≥ 50% contact between hardware and countersink”. Three sources, three numbers. The cells above say which criterion governs at each thickness, and the answer is instructive: on thin sheet it is the LAND IN MILLIMETRES, not any percentage. In 0.8 mm aluminium the 90% rule would allow 0.72 mm and the 0.25 mm land allows only 0.55, so the land wins. The percentage is the rule people quote; the land is the rule that bites. These dimensions come from a published standard’s table, not from a formula. The standard itself is cited below and the printed values are attributed to the catalogue they were taken from; a different publisher may round differently in the last digit.
Material	Published % of thickness	Minimum land (mm)	Max depth in 0.8 mm	in 1.0 mm	in 1.5 mm	in 2.0 mm	in 3.0 mm
Aluminium	90%	0.25	0.55 (land)	0.75 (land)	1.25 (land)	1.75 (land)	2.70 (%)
Steel	80%	0.25	0.55 (land)	0.75 (land)	1.20 (%)	1.60 (%)	2.40 (%)
Stainless steel	66%	0.55	0.25 (land)	0.45 (land)	0.95 (land)	1.32 (%)	1.98 (%)
A job shop’s blanket rule	60%	—	0.48 (%)	0.60 (%)	0.90 (%)	1.20 (%)	1.80 (%)

The recess is ours; the drill is not

Stated plainly because a reader arriving here may have wanted the other page. Converters owns the size table and the drill; mechanical owns the feature on the part. A tap drill diameter is a conversion. The depth and diameter of the counterbore that swallows a cap screw head is a machining feature, and that is this page.
What	Where it lives
Tap drill sizes, clearance-hole drill sizes, and the metric, fractional, number and letter drill series	The converters plugin owns all of it. That is a size-table conversion: you have a thread or a hole and you want the drill that makes it.
The counterbore’s diameter and depth, the countersink’s top diameter and its angle, the land left under either, and what happens when the angle is wrong	Here. These are features on the part, with dimensions that go onto a drawing and a tolerance that has to be met. The drill is the tool; the recess is the part.
The clearance hole DIAMETER itself	Both, for different reasons. ISO 273’s fine, medium and coarse classes are printed here because a countersink’s depth is measured from where the cone meets the hole, so the page cannot compute anything without them. Which DRILL cuts a 6.6 mm hole is a converter question.
Socket head, flat head, button head and shoulder screw dimensions	Each has its own page on this site already. This page takes their head dimensions as inputs and computes the recess.
Preload, torque and the bearing stress under the head	The bolted-joint cluster. What this page gives it is the bearing AREA, which is the geometry half of that calculation — and the area is what the below-flush setting changes.

Two angles, two errors, and the rule that actually limits the depth

The included angle is a standards fork with a physical consequence, and the best thing about this page is that the angles do not have to be taken on trust. 90° is the metric norm — ISO 10642, DIN 7991 — and 82° ± 2° is the inch norm, ASME B18.3. Both fall straight out of the standards’ own dimension tables. ISO 10642’s published head height is exactly (dk − d)/2 at every size from M3 to M20, and (dk − d)/2 is the height of a cone with a 45° half angle and of nothing else. ASME B18.3’s published flat-head heights are exactly (dk − d)/(2 tan 41°) at all thirteen of its sizes, to the last printed digit, and that is an 82° cone. Two standards, two arithmetic checks, no drawing notes required.

What happens when they are mixed is not symmetric, and the dangerous case is the one that looks right. Put an 82° head into a 90° countersink cut to the same top diameter and it cannot descend: an 82° flank falls away more slowly than a 90° bore narrows, so the head’s lower edge jams against the cone and the head stands proud. On an M6 that is 0.51 mm, and cutting the 90° cone deeper does not help — the head simply goes further in without ever touching the flank. Now reverse it. A 90° head in an 82° cone of the same top diameter sits down flush, and bears on a LINE at the rim of the countersink rather than on the cone at all. It looks perfect and it is the worst bearing condition available to a screw: the whole preload through a circle of contact instead of a conical band. That asymmetry was worked out here from the cone geometry and then checked by brute force — sampling both profiles at twenty thousand radii and finding the deepest the head can go without the profiles crossing — at five geometries, agreeing with the closed form to a ten-thousandth of a millimetre.

The counterbore side is simpler and has two facts worth memorising. An ISO 4762 socket head cap screw’s head height is EXACTLY the nominal diameter at every size, and its head diameter is exactly 1.5 d from M12 up. So a counterbore for a socket head is 1.5 d across and d deep, plus clearance — you can size one without a table. Two published sources agree on every counterbore diameter and differ on the depth, 6.5 mm against 6.8 at M6, because the depth is the head height plus an allowance each publisher chose; both are on this page as separate columns. Whole-Spec’s own warning is worth repeating: making the counterbore diameter exactly the head diameter “is a mistake”, because plating, the fillet under the head and the tool’s runout all need room.

The sheet-thickness limit is not one number, and the number people quote is not the one that binds. This page was briefed with a rule of about 70% of the sheet. No source for that was found. What is published is material-dependent and comes with a second criterion: PASS Stanztechnik give 90% of thickness for aluminium, 80% for steel and 66% for stainless, each with a minimum cylindrical land — 0.25 mm generally and 0.55 mm for stainless — and SendCutSend, a job shop, publish a single stricter 60%. The table on this page evaluates all of them at five thicknesses and marks which criterion governs, and the result is the useful bit: on thin sheet it is the LAND IN MILLIMETRES, not any percentage. In 0.8 mm aluminium the 90% rule allows 0.72 mm and the 0.25 mm land allows only 0.55, so the land wins by a wide margin. Below the land you have a knife edge, which will not bear and may not survive being cut.

And the boundary, because a reader may have wanted a different page. Drill sizes — tap drill, clearance drill, metric, fractional, number and letter series — are a size-table conversion and live in the converters plugin. This page owns the feature on the part: the pocket, the cone, the land underneath and the tolerance window between too shallow and too deep. ISO 273’s clearance-hole classes appear here because a countersink’s depth is measured from where the cone meets the hole, so nothing can be computed without them. For the fastener at the other end, the hex nut dimensions calculator; for the head’s own bearing face when it is NOT recessed, the hex head bolt dimensions calculator; for the washer that spreads what the head cannot, the washer dimensions calculator; and when the hole is there to LOCATE rather than to clamp, the dowel pin and hole calculator, because a screw in a clearance hole locates nothing.

Frequently asked questions

Is a countersink 82° or 90°?

90° for metric screws (ISO 10642, DIN 7991) and 82° ± 2° for inch ones (ASME B18.3). You do not have to trust that: both angles are recoverable from the standards’ own dimension tables. ISO 10642’s head height is exactly (dk − d)/2 at every size, which is a 45° half angle; ASME B18.3’s is exactly (dk − d)/(2 tan 41°) at every size, which is 41°. Aerospace sheet work also uses 100°, and 60° is a centre-drill and deburring angle rather than a screw seat.

What happens if I use a screw in the wrong-angle countersink?

It depends which way round, and only one of the two announces itself. An 82° head in a 90° recess stands PROUD — half a millimetre on an M6 — because the bore narrows faster than the head does, so the head’s lower edge jams. Cutting the recess deeper does not fix it. A 90° head in an 82° recess sits flush and looks correct, but bears on a LINE at the rim of the countersink instead of on the cone, so the whole preload goes through a circle of contact. That is the failure to worry about: it passes visual inspection.

How deep should a counterbore be?

The head height plus however far below flush you want the head. For an ISO 4762 socket head cap screw the head height is exactly the nominal diameter — 6 mm on an M6, 10 on an M10 — so flush is d deep and a typical working depth is d + 0.5 mm. The two published tables this page cites differ only in that allowance: 6.5 mm against 6.8 mm at M6. Check what is left underneath: that land carries the screw’s bearing load into the part.

What counterbore diameter should I use for a socket head cap screw?

The head diameter plus clearance, and the head diameter is exactly 1.5 d from M12 up (a little wider below). The published figures add 1 mm of diametral clearance up to M8, 1.5 at M10, 2 at M12 and M16 and 3 at M20. Do not make it exactly the head diameter — Whole-Spec call that “a mistake”, and the reasons are plating thickness, the fillet under the head and the counterbore tool’s own runout.

How deep can I countersink a thin sheet?

It depends on the material, and the binding rule on thin sheet is not a percentage. PASS Stanztechnik publish 90% of thickness for aluminium, 80% for steel and 66% for stainless, each with a minimum cylindrical land of 0.25 mm (0.55 for stainless); SendCutSend publish a blanket 60%. Evaluate both criteria and on thin sheet the LAND wins: in 0.8 mm aluminium the 90% rule allows 0.72 mm of depth and the 0.25 mm land allows only 0.55. This page computes both and says which governs. Below the land you have a knife edge that will not bear.

Why is the countersunk head’s theoretical diameter bigger than the actual head?

Because the head is truncated. ISO 10642 prints a theoretical maximum head diameter — the apex-to-apex cone — and an actual minimum head diameter well below it, because the sharp edge is removed. The RECESS has to match the theoretical cone, because that is the cone the flank lies on; the actual diameter tells you what you will measure on the part. At M6 those are 13.44 mm and 11.34 mm, which is a 2 mm difference — enough to size a countersink wrong if you use the measured head.

Does this page give the drill size?

No. Drill sizes — tap drill, clearance drill, metric, fractional, number and letter series — are a size-table conversion and live in the converters plugin. This page computes the FEATURE: the counterbore’s diameter and depth, the countersink’s top diameter and depth, the material left underneath and the window between cutting too shallow and too deep. It does print ISO 273’s clearance-hole diameters, because a countersink’s depth is measured from where the cone meets the hole and nothing can be computed without them.

Related calculators

References

ISO 10642, Hexagon socket countersunk head screws (DIN 7991), with the dimensions read from Fuller Fasteners’ specification page. The 90° included angle is not taken on trust: the standard’s own head height k is exactly (dk − d)/2 at every size from M3 to M20, which is a 45° half-angle and nothing else. Its theoretical head diameter is exactly 2.24 d from M3 to M12.
ASME B18.3, Socket Cap, Shoulder, Screw Head and Hex Keys (Inch Series), with the flat-head dimensions read from Fuller Fasteners’ specification page, which prints the included angle as 82° ± 2°. Again checked rather than assumed: the published head height is exactly (dk − d)/(2 tan 41°) at all thirteen sizes, to the last printed digit.
ISO 4762 (DIN 912), Hexagon socket head cap screws, from Fuller Fasteners’ specification page. Its head height is exactly the nominal diameter at every size, and its head diameter is exactly 1.5 d from M12 up.
Counterbore diameters and depths from two independent publishers: Whole-Spec’s Counterbore Spec Table M3~M20 (ISO 4762) and The Engineer’s Bible’s Counterbore Hole Size for Socket Head (ISO 4762). They agree on every DIAMETER and differ on the DEPTH — 6.5 mm against 6.8 mm at M6 — because the depth is the head height plus an allowance each publisher chose. Whole-Spec also note that making the counterbore diameter exactly the head diameter “is a mistake”.
ISO 273, Fasteners — Clearance holes for bolts and screws. Cited by number; the fine / medium / coarse columns used here are taken from EKINSUN’s What Size Clearance Hole for a Metric Bolt? ISO 273 Chart, which labels them H12, H13 and H14.
PASS Stanztechnik AG. Max. depth of countersinks. Published maxima for a 90° countersink as a fraction of sheet thickness: aluminium “max. 90% of s”, steel “max. 80% of s”, stainless steel “max 66% of s”, each with a minimum cylindrical land — “min. 0,25 mm zyl. part or depth”, and 0.55 mm for stainless. SendCutSend’s own countersinking page gives a single stricter figure, “should not exceed 60% of material thickness”, with a 0.125 in minimum material thickness and a target of “≥ 50% contact between hardware and countersink”. Three published limits, three different numbers, and the binding one on thin sheet is the land in millimetres rather than any of the percentages.