Torx and Drive Size Calculator
Torx and Drive Size Calculator
Torx, Phillips, Pozidriv, slotted and square side by side — with the ISO socket dimension and the driver point diameter kept apart, the external E series shown not to be the T series, and the drive size asked by head style because that is what decides it.
Torx and drive sizes
A #10 socket cap screw, with a T27 and an E10 looked up beside it
A lookup, two standards, and one rule that is not a table
- A
- ISO 10664’s dimension across the lobes of the SOCKET — the hole in the screw. 1.75 mm at size 6 and 22.40 at size 100. The standard does not control it with a tolerance: it controls the recess with a GO and a NOT GO gauge, and at size 6 those are 1.695–1.709 and 1.778–1.785 mm
- d_driver
- the point diameter of the TOOL, which is the number most published Torx charts are actually printing. It is 1 to 4 per cent below A, because it has to go in. Two distributors’ catalogues agree on it to 0.03 mm
- 0.060 + 0.013 n
- the American screw gauge rule, in inches. It is a rule and not a table, and it reproduces every printed gauge diameter exactly: #6 is 0.138 in, #10 is 0.190 in. A #10 is 4.826 mm, which is 3.6 per cent under an M5
- type H, type Z
- ISO 4757’s two cross recesses — Phillips and Pozidriv. The standard defines the RECESS; ISO 8764-1 is a separate standard for the driver tip that goes into it, and DIN 5260 is the German blade form. Naming the wrong one describes the wrong object
- head style
- the variable this page exists to put in. The recess must fit inside the head, so a grub screw and a socket cap screw of the same thread take Torx sizes several steps apart. Any chart with one Torx column has silently chosen a head style for you
Worked example
A #10 socket cap screw, with a T27 and an E10 looked up beside it
THE SCREW. A #10 gauge is 0.060 + 0.013 × 10 = 0.190 inches, which is 4.826 mm. Not an M5: an M5 is 5.000, so the two are 3.6 per cent apart on major diameter, which is enough that an M5 will not enter a #10 nut
THE DRIVE. From Wiha's chart, a #10 SOCKET CAP screw takes a T27. Change nothing but the head style and the answer moves a long way: the same #10 thread as a grub screw takes a T10, as a pan head a T25 and as a flat head a T20. The recess has to fit inside the head, and a socket cap head is one and a half to one and three quarter times the thread diameter while a grub screw has no head at all
THE DIMENSION. A T27 is 5.10 mm across the lobes. Note two things about that size. It is not in ISO 10664:1999 at all — the standard's table jumps from 25 to 30 — and yet it is one of the commonest sizes in a modern car. And the recess is 1.057 times the screw's own diameter, which is why the drive is the strong part and the thread is not
THE TOOL IS NOT THE HOLE. Two distributors print the T27 driver point diameter as 4.99 mm. That is 0.11 mm smaller than the socket, or 2.2 per cent. It has to be, or the bit would not go in. Almost every Torx chart on the web is one of these two columns and does not say which, and the gap between them is larger than the gauge window the standard allows on the recess
AND AN E10 IS NOT A T10. The external E series is a different family: the lobes are on the OUTSIDE of the head and a socket goes over it. An E10 is 9.30 mm across the lobes against a T10's 2.80 mm — 3.32 times the size, for the same digits. An E10 is used on M8 screws; a T10 on #4 to #5. If someone hands you an E-series socket set expecting it to cover T sizes, it will not cover any of them
WHAT TO DO WITH THE OTHER DRIVES. The same #10 screw takes a PH3 Phillips or a PZ3 Pozidriv — the same number, and they are NOT the same recess and do not interchange. Slotted, it takes a 7.9 mm blade. Square, a #2. The table on this page runs the full cross-reference, and hex and Allen sizes are covered elsewhere on this site rather than here
The two Torx tables, side by side
| Size | A, across the lobes (mm) | B (mm) | Driver point diameter (mm) | Clearance (mm) | Clearance as % of A | In ISO 10664:1999? |
|---|---|---|---|---|---|---|
| T6 | 1.750 | 1.270 | 1.685 | 0.065 | 3.71 | yes |
| T7 | 2.100 | — | 1.995 | 0.105 | 5.00 | no — manufacturer size |
| T8 | 2.400 | 1.750 | 2.305 | 0.095 | 3.96 | yes |
| T9 | 2.600 | — | 2.495 | 0.105 | 4.04 | no — manufacturer size |
| T10 | 2.800 | 2.050 | 2.730 | 0.070 | 2.50 | yes |
| T15 | 3.350 | 2.400 | 3.265 | 0.085 | 2.54 | yes |
| T20 | 3.950 | 2.850 | 3.860 | 0.090 | 2.28 | yes |
| T25 | 4.500 | 3.250 | 4.430 | 0.070 | 1.56 | yes |
| T27 | 5.100 | — | 4.990 | 0.110 | 2.16 | no — manufacturer size |
| T30 | 5.600 | 4.050 | 5.520 | 0.080 | 1.43 | yes |
| T35 | 5.900 | — | — | — | — | no — manufacturer size |
| T40 | 6.750 | 4.850 | 6.650 | 0.100 | 1.48 | yes |
| T45 | 7.930 | 5.640 | 7.820 | 0.110 | 1.39 | yes |
| T50 | 8.950 | 6.450 | 8.830 | 0.120 | 1.34 | yes |
| T55 | 11.350 | 8.050 | 11.220 | 0.130 | 1.15 | yes |
| T60 | 13.450 | 9.600 | 13.250 | 0.200 | 1.49 | yes |
| T70 | 15.700 | 11.200 | 15.510 | 0.190 | 1.21 | yes |
| T80 | 17.750 | 12.800 | — | — | — | yes |
| T90 | 20.200 | 14.400 | — | — | — | yes |
| T100 | 22.400 | 16.000 | — | — | — | yes |
Every drive that fits a given screw
| Screw | Major dia (mm) | Torx, set screw | Torx, socket cap | Torx, pan head | Torx, flat head | Phillips | Pozidriv | Slotted blade | Square |
|---|---|---|---|---|---|---|---|---|---|
| #2 | 2.18 | T3 | — | T7 | T6 | PH1 | PZ1 | 2.4 mm | — |
| #3 | 2.51 | T5 | — | T8 | T7 | PH1 | PZ1 | 3.2 mm | — |
| #4 | 2.84 | T6 | T10 | T9 | T8 | PH1 | PZ1 | 3.2 mm | — |
| #5 | 3.17 | T7 | T10 | T10 | T9 | PH2 | PZ2 | 4.0 mm | #1 |
| #6 | 3.51 | T7 | T15 | T15 | T10 | PH2 | PZ2 | 4.8 mm | #1 |
| #8 | 4.17 | T8 | T25 | T20 | T15 | PH2 | PZ2 | 7.9 mm | #2 |
| #10 | 4.83 | T10 | T27 | T25 | T20 | PH3 | PZ3 | 7.9 mm | #2 |
| #12 | 5.49 | T10 | T27 | T27 | T25 | PH3 | PZ3 | 9.5 mm | #3 |
| 1/4 | 6.35 | T20 | T30 | T30 | T27 | PH3 | PZ3 | 9.5 mm | #3 |
| 5/16 | 7.94 | T27 | T45 | — | T40 | PH4 | PZ4 | 11.1 mm | — |
| 3/8 | 9.52 | T30 | T50 | T45 | T40 | PH4 | PZ4 | 11.1 mm | — |
| 7/16 | 11.11 | T40 | T55 | T50 | T50 | PH4 | PZ4 | 12.7 mm | — |
| 1/2 | 12.70 | T45 | T55 | T55 | T55 | PH4 | PZ4 | 12.7 mm | — |
External Torx, and why an E10 is not a T10
| External size | Across the lobes (mm) | Inch screw | Metric screw | Same-numbered internal size | Its dimension (mm) | E ÷ T |
|---|---|---|---|---|---|---|
| E4 | 3.80 | #6 | M3 | T4 | 1.35 | 2.81× |
| E5 | 4.70 | #8 | M4 | T5 | 1.50 | 3.13× |
| E6 | 5.60 | #10 | M5 | T6 | 1.75 | 3.20× |
| E7 | 6.10 | — | — | T7 | 2.10 | 2.90× |
| E8 | 7.40 | 1/4 in | M6, M7 | T8 | 2.40 | 3.08× |
| E10 | 9.30 | 5/16 in | M8 | T10 | 2.80 | 3.32× |
| E12 | 11.10 | 3/8 in | M10, M11 | no T12 exists | — | — |
| E14 | 12.80 | 7/16 in | M12 | no T14 exists | — | — |
| E16 | 14.70 | 1/2 in | — | no T16 exists | — | — |
| E18 | 16.60 | 9/16 in | M14 | no T18 exists | — | — |
| E20 | 18.40 | 5/8 in | M16 | T20 | 3.95 | 4.66× |
| E24 | 22.10 | 3/4 in | M18, M20 | no T24 exists | — | — |
Phillips against Pozidriv against Torx
| Phillips (ISO 4757 type H) | Pozidriv (ISO 4757 type Z) | Torx (ISO 10664) | |
|---|---|---|---|
| The recess | Four radial slots, tapered flanks, rounded corners, a pointed centre | The same four slots PLUS four smaller ones at 45°, parallel flanks, a blunt tip | Six lobes, near-radial contact faces, no taper at all |
| The visual tell | Just the cross | Four short tick marks between the arms of the cross, at 45°. If you can see them it is Pozidriv | A six-pointed star |
| Contact angle | The flank is tapered, so the driving force has a component pushing the bit OUT of the recess | The flanks are parallel, so much less of the force pushes the bit out | The faces are nearly radial, so essentially none of it does |
| Cam-out | Yes, and at a torque that depends on how hard you push. It is the practical limit on a Phillips joint | Much less than Phillips, which is what it was changed for | By design, no. The bit does not climb out; it twists off or the screw fails first |
| Wrong bit in it | A Pozidriv bit in a Phillips screw touches on the wrong flanks, sits proud and cams out early | A Phillips bit in a Pozidriv screw fits loosely, contacts on a small part of the flank and rounds the recess | A Torx Plus driver will not enter; a Torx driver in a Torx Plus screw is loose and must not be taken to full torque |
| Torque for a given head size | Lowest of the three | Higher than Phillips | Highest by a wide margin, which is why it took over in volume assembly |
| Sizes | PH0 to PH4 | PZ0 to PZ5 | T1 to T100 internally, E4 to E24 externally, IP and EP for Torx Plus |
Two Torx tables, two cross recesses that will not interchange, and the head style that decides the size
Hex and Allen sizes are covered elsewhere on this site. This page is about the other drive systems — the hexalobular Torx family, the two cross recesses, slotted and square — and about the cross-reference between them, which is the thing that is hard to find and easy to get wrong.
There are two Torx tables in circulation and hardly anybody says which one they are printing. ISO 10664 gives A, the dimension across the lobes of the SOCKET: 1.75 mm at size 6, 4.50 at 25, 22.40 at 100. Tool catalogues give the point diameter of the DRIVER, which is a different measurement of a different object and runs 1 to 4 per cent smaller, because the bit has to go into the screw. Two distributors’ driver tables agree with each other to 0.03 mm and disagree with ISO’s socket figure by up to 0.15 — more than the GO/NOT GO window the standard allows on the recess itself, so the discrepancy is not a tolerance and cannot be split. Both columns are on this page, labelled, with the clearance between them computed. If you are checking a recess, use A. If you are checking a bit, use the driver column. If a chart does not say, it is probably the driver.
Phillips and Pozidriv are not interchangeable, and this is the most useful paragraph on the page. ISO 4757 defines exactly two cross recesses: type H, which is Phillips, and type Z, which is Pozidriv. They use the same size numbers, 0 to 4, and they fit each other badly in both directions. A Phillips driver in a Pozidriv screw is loose: it contacts on a small part of the flank, it cams out early, and it rounds the recess. A Pozidriv driver in a Phillips screw contacts on the wrong flanks, sits proud, and does the same thing faster. The tell is on the SCREW and it takes a second to learn: a Pozidriv head carries four short tick marks at 45° to the arms of the cross, stamped by the punch into the flat of the head. A Phillips head has nothing but the cross. If the marks are there, use a PZ bit; if they are not, use a PH one. The failure mode when you get it wrong is not that the screw will not turn — it turns, it feels tight, and the recess is being destroyed while it does.
Cam-out, and one story that will not go away. A Phillips recess has tapered flanks, so part of the torque you apply becomes an axial force pushing the bit out of the screw; past a certain torque, at a given down-force, it climbs out. Pozidriv’s parallel flanks reduce that and Torx’s near-radial faces essentially eliminate it, which is why a Torx head half the size of a Phillips one will take more torque, and why volume assembly moved to it. That much is geometry and is not in dispute. What IS in dispute is the story that Phillips was deliberately designed to cam out so that assembly lines could not overtighten: there is no good evidence for it and it does not appear in the original patents. Cam-out is a consequence of the shape, whoever intended it.
The size that fits your screw depends on the head, not just the thread. This is the finding that makes a single-column cross-reference chart misleading. The recess has to fit inside the head with enough wall left to carry the torque, so the same thread takes very different drives in different head styles: a #8 screw is a T8 as a grub screw, a T15 as a flat head, a T20 as a pan head and a T25 as a socket cap screw. Four sizes, one thread. The table on this page carries four Torx columns for that reason, and the selector at the top asks for the head style before it answers. For a metric reader converting, the gauge rule is 0.060 + 0.013n inches: #6 is 3.51 mm, #8 is 4.17, #10 is 4.83 and #12 is 5.49, so the usual pairings with M3.5, M4, M5 and M5.5 are near misses rather than equivalences. How near is a different page’s subject on this site.
Frequently asked questions
How big is a T25 across the lobes?
4.50 mm as a socket, which is ISO 10664’s nominal A dimension, and about 4.43 mm as a driver point, which is what two distributors’ catalogues print. Both figures are correct and they are measurements of different objects — the hole in the screw and the tool that goes into it. The 0.07 mm between them is the clearance. ISO 10664 does not put a tolerance on A at all: it controls the recess with a GO and a NOT GO gauge, which at the smallest size are 1.695–1.709 mm and 1.778–1.785 mm around a 1.75 nominal.
Is an E10 the same as a T10?
No, and they are not even close. An E10 is an EXTERNAL Torx size — the lobes are on the outside of the head and a socket goes over it — and it measures 9.30 mm across the lobes. A T10 is an internal recess and measures 2.80 mm. The E number is roughly 3.3 times the T number of the same digits, and the ratio is not constant, so there is no conversion to learn. An E10 is used on M8 screws; a T10 on #4 to #5. An E-series socket set will not fit any T-series screw.
What Torx size fits an M5 screw?
It depends on the head. A metric socket cap screw around M5 — which is close to a #10 — typically takes a T27; a pan head T25; a flat head T20; and a grub screw T10. That four-size spread is the reason this page asks for the head style. Note also that M5 and #10 are not the same screw: #10 is 4.826 mm and M5 is 5.000, which is 3.6 per cent apart and enough to stop one entering the other’s nut.
How do I tell a Pozidriv screw from a Phillips one?
Look for four short tick marks in the flat of the head, at 45° to the arms of the cross. They are stamped by the Pozidriv punch and a Phillips head does not have them. On the bit rather than the screw, a Pozidriv driver has parallel flanks, a blunt tip and four extra ribs between the main blades; a Phillips driver has tapered flanks, a pointed tip and rounded corners. Getting it wrong does not stop the screw turning, which is exactly why it does damage: the screw goes in feeling tight while the recess is being chewed out.
Why does Torx take more torque than Phillips?
Because of where the driving faces point. A Phillips flank is tapered, so the force you put into it has a component along the screw axis pushing the bit out; past some torque, at whatever down-force you are applying, it climbs out and the recess takes the damage. A Torx lobe face is nearly radial, so almost all of the force goes into turning and essentially none into lifting the bit. The practical consequence is that a smaller Torx head carries more torque than a larger Phillips one, which is why volume assembly moved to it and why a Torx joint fails by twisting the bit off rather than by rounding the screw.
Are all Torx sizes in the ISO standard?
No. ISO 10664:1999 tabulates sixteen sizes — 6, 8, 10, 15, 20, 25, 30, 40, 45, 50, 55, 60, 70, 80, 90 and 100 — and designates them by socket number rather than with a T. T1 to T5, T7, T9, T27 and T35 are manufacturer sizes outside that table, which does not make them rare: T27 and T35 are among the commonest sizes in a car and T7 and T9 are everywhere in consumer electronics. Their dimensions come from tool makers rather than from a standards body, so where a dimension matters, measure the recess.
What about Torx Plus and security Torx?
Torx Plus is a later family with a squarer lobe for more torque and less wear, designated IP internally and EP externally. The compatibility runs one way: a standard Torx driver will enter a Torx Plus recess but fits loosely and should not be taken to full torque in it, while a Torx Plus driver will not enter a standard Torx recess at all. Security Torx is the same recess with a pin in the centre, which blocks a solid driver; the matching bits are hollow and are sold freely, so it is a deterrent rather than a security measure.
Related calculators
References
- ISO 10664:1999, Hexalobular internal driving feature for bolts and screws. Copyrighted and cited by number. The sixteen nominal A and B values used here were read from a published copy of the standard and then checked, row by row, against Wikipedia’s Torx article, which prints the same sixteen A values independently. Two points from the standard itself are worth stating and are not tables: it designates its sizes by a plain SOCKET NUMBER rather than by “T”, and it controls the recess with GO and NOT GO gauges rather than with a tolerance on A — for socket No. 6 the gauge limits are 1.695 to 1.709 mm GO and 1.778 to 1.785 mm NOT GO, which brackets the 1.75 nominal.
- ISO 4757:1983, Cross recesses for screws. Cited by number. It defines exactly TWO cross recesses — type H, which is the Phillips form, and type Z, which is the Pozidriv form — in sizes 0 to 4, with penetration gauging for both. It is the recess standard; ISO 8764-1:2004, Assembly tools for screws and nuts — Screwdrivers for cross-recessed head screws — Part 1: Driver tips, is the separate standard for the tool that goes into it, and DIN 5260 is the German blade-form standard. Three different documents, and a page that names the wrong one is describing the wrong object.
- Intafast Ltd, Torx / Hexalobular guide, and Compel Group, Torx Size Chart (both read 29 September 2026). Two independent distributors’ tables of the DRIVER point diameter, which is a different number from ISO 10664’s socket A dimension and is the one most “Torx size charts” are actually printing. They agree with each other to 0.03 mm at every shared size (T6: 1.67 and 1.70; T10: 2.72 and 2.74; T15: 3.26 and 3.27) and sit consistently 1 to 4 per cent below the socket nominal — which is the clearance that lets the driver in, and is why the two tables must never be mixed.
- Wikipedia, Torx (read 29 September 2026). Used for three things this batch could not get from the standard: the manufacturer sizes ISO 10664:1999 does not carry (T1–T5, T7, T9, T27, T35), the external E-series across-lobes dimensions, and the plain statement that “the external ‘E’ Torx nominal sizing does not correlate to the ‘T’ size”. Its internal A column agrees with the published ISO 10664 table at all sixteen shared sizes, which is what qualifies it as the second source for that column.
- Wikipedia, List of screw drives (read 29 September 2026). The source for the Phillips and Pozidriv size-to-screw ranges, for the ISO 2380 slotted blade widths, and for the geometric difference between the two cross recesses: a Phillips driver has “slightly tapered flanks, a pointed tip, and rounded corners”, a Pozidriv driver has “parallel flanks, a blunt tip, and additional smaller ribs at 45° to the main slots”. It is also the source for the correction on this page’s cam-out paragraph: the story that Phillips was DESIGNED to cam out has “no good evidence” behind it and “is not mentioned in the original patents”.
- Wiha Tools, Torx Screw Size Chart (read 29 September 2026). The source for the screw-size to Torx-size cross-reference, and for the finding that makes a one-column chart useless: the Torx size depends on the HEAD STYLE as much as on the diameter. Wiha prints nine head styles, and a #8 screw takes a T8 as a socket set screw and a T25 as a socket cap screw — four sizes apart on the same thread.
- Accu Ltd, Pozidriv vs. Phillips, what’s the difference? (read 29 September 2026). The practical statement of the mismatch: a Phillips driver “may appear to fit into a Pozi screw” but the cam-out risk is much higher, and the visual tell is that a Pozidriv recess has eight radial slots — the four of the cross plus a secondary set of four at 45° — where a Phillips has four.
- EngineerFix, Square Bit Size Chart (read 29 September 2026). The source for the square (Robertson) size to screw gauge mapping and its colour code: #0 yellow for gauge 3 and 4, #1 green for 5 to 7, #2 red for 8 to 10, #3 black for 12 and 14. It also says the thing that stops this page printing a dimension for them: “actual tip dimensions vary slightly between manufacturers”, and there is no single published across-flats to quote.
