The inputs, in full
Stress areas for the metric sizes are the published proof load divided by the published proof stress for the class - arithmetic on two published figures rather than a separate measurement. For reference the ISO formula is As = 0.7854 x (d - 0.9382 x P)^2, with d the nominal diameter and P the pitch.
Estimate by lubrication condition
Computed here from the equation and the inputs above, not copied from a chart.
T = K x D x P. D is the nominal diameter and P is the target preload, taken here as 75 percent of the proof load. K is the nut factor: a single empirical number standing in for the friction in the threads and under the head, and it is where every disagreement between torque charts comes from. Nothing on these pages is a torque measured on a joint - the published columns are what named charts printed, and the estimate columns are this equation with the K, D and P shown beside the answer.
What the published charts print
Verbatim, with each chart's own stated assumptions beside it. 3 charts publish a figure for this size.
Why no single number is the answer
Almost all of the torque you apply is spent on friction: roughly half under the head and half in the threads, with what is left going into stretching the bolt. K is the one number standing in for all of that, and it is measured rather than derived. Two charts that both say "ISO property class 8.8 M18" can differ by a third purely because one assumed a dry fastener and the other a lubricated one.
A torque figure is an estimate of preload, not a measurement of it. Even with the right K, torque control typically lands the preload within about plus or minus 25 to 30 percent, because almost all of the applied torque goes into friction rather than stretch. Where the preload actually matters, measure it - turn of the nut, bolt elongation, a load-indicating washer or a tensioner - rather than trusting a wrench setting. And a torque table cannot know what your joint is made of: it assumes the threads and the clamped material can carry the preload the bolt can.
The preload these figures target comes from the grade, not the size alone. ISO property class 8.8 is medium carbon steel or carbon steel with additives, quenched and tempered, and its proof load is banded by diameter - which is why a torque chart that gives one figure per grade is wrong for part of the range.
Frequently asked questions
What is the torque for a ISO property class 8.8 M18 bolt?
Dry and as received, the estimate is 311 N-m (229 lb-ft). That is T = K x D x P with K = 0.2, D = 18 mm and P = 86.4 kN, which is 75 percent of the published proof load of 115.2 kN. Change any of the three and the answer changes.
Does lubricating the threads change the torque?
Yes, substantially. At the same preload, a lubricated M18 needs 233 N-m (172 lb-ft) against 311 N-m (229 lb-ft) dry - the nut factor drops from 0.2 to 0.15. Using the dry figure on a lubricated bolt overloads it; using the lubricated figure on a dry one leaves the joint loose.
What clamp load does this torque produce?
86.4 kN, the target preload these figures are built around. It is 75 percent of the proof load, which is 115.2 kN for a ISO property class 8.8 M18. Every chart cross-checked here targets 75 percent of proof load, so the disagreements between them are about K alone. A chart that targets 65 percent, as some do for reusable joints, will give figures about 13 percent lower for the same fastener and the same K. Always read the clamp fraction before comparing two charts.
How accurate is a torque wrench setting?
A torque figure is an estimate of preload, not a measurement of it. Even with the right K, torque control typically lands the preload within about plus or minus 25 to 30 percent, because almost all of the applied torque goes into friction rather than stretch. Where the preload actually matters, measure it - turn of the nut, bolt elongation, a load-indicating washer or a tensioner - rather than trusting a wrench setting. And a torque table cannot know what your joint is made of: it assumes the threads and the clamped material can carry the preload the bolt can.
Where these figures come from
Dimensions are as defined by ISO 898-1. This site is an independent reference and is not affiliated with, endorsed by or approved by any standards body.
Every value on this page was cross-checked against 6 independent published sources:
- Torque-tension relationship, ISO 898-1 property class 4.6, 8.8, 10.9 and 12.9, metric coarse thread - Fastenal, Jan 2016 (read 2026-08-27)
- Bolt torque chart: SAE and metric specs by grade - WorkshopCalc, edition not stated (read 2026-08-27)
- Torque chart for class 8.8, 10.9 and 12.9 metric fasteners - Ecom Fasteners, edition not stated (read 2026-08-27)
- EN ISO 898-1 (2013) Table 3, stress under proof load by property class - Sidex s.r.l., EN ISO 898-1:2013 (read 2026-08-27)
- ISO 898 Part 1 - 2013 (extract), Table 3 - Kova Fasteners, ISO 898-1:2013 (read 2026-08-27)
- Mechanical properties, screws and studs, EN ISO 898-1:2013 - Nordic Fastening Group, EN ISO 898-1:2013 (read 2026-08-27)
Sources disagree here. For lubricated the published charts give 233.0 N-m (torque-fastenal-metric), 184 N-m (torque-ecomfasteners), 217 N-m (torque-workshopcalc). They differ because they assume different nut factors, not because the fastener changes; the K each one states, or the K its own numbers imply, is printed beside it. For plain and dry the published charts give 311.0 N-m (torque-fastenal-metric), 245 N-m (torque-ecomfasteners), 290 N-m (torque-workshopcalc). They differ because they assume different nut factors, not because the fastener changes; the K each one states, or the K its own numbers imply, is printed beside it.
The nut factor for a zinc-plated, dry fastener: K = 0.17 (torque-fastenal-inch) against K = 0.18 (torque-engineersedge). Both are printed. K is an empirical lumped friction term, not a property of the coating, and a six percent spread between two careful charts is a fair picture of how well it is known. The estimate columns use 0.17 and show the range.
The nut factor two of the charts never state: states K for every column (torque-fastenal-inch) against states no K (torque-workshopcalc) against states no K (torque-ecomfasteners). Where a chart prints a torque without saying what nut factor produced it, the K its own numbers imply is computed from its published clamp load and printed beside the figure. That is arithmetic on what the chart published, and it is labelled as implied rather than stated.
Proof load: the standard's table against the charts' arithmetic: 1/2-13 Grade 5 proof load 12,100 lb (j429-1999) against clamp load 9,046 lb, which is 75 percent of 12,061 lb (torque-fastenal-inch). SAE J429 Table 5 rounds its proof loads; the torque charts multiply the stress area by the proof stress and keep the unrounded product. The gap is about a third of a percent. The proof load printed on these pages is the published one from the standard, and the published torque columns are left exactly as their charts printed them rather than rescaled to match.