Bike Bolt Torque Guide

Heads up: we earn a commission if you buy through the links on this page, at no extra cost to you. It never decides what makes the list — here’s how that works.

Torque is not tightness — it is a measurement of the twisting force you applied to get there.What actually holds a component in place is the clamping load (“preload”) in the stretched bolt, and torque is the only practical proxy a home mechanic has for it. Too little and the part slips, creaks, or fatigues the bolt through repeated micro-movement. Too much and you strip the thread, deform the component, or crack a carbon part that was perfectly fine ten seconds earlier.

The figures below are all published by the people who make the parts. We have not derived, averaged or estimated a single one, and Park Tool’s own caveat on its reference table applies to this page too: “Numbers provided for reference only. Note that some companies do not specify torque for certain components or parts. Contact the manufacturer for the most up to date specifications.”

The rule that outranks every table

If a figure is printed on your component, that figure wins.Modern stems, seatpost clamps and carbon bars usually have the spec laser-etched next to the bolt, and component manufacturers publish per-model figures in their dealer manuals — Shimano specifies 5–7 N·m for a front derailleur clamp bolt in its own dealer’s manual, for instance. A generic table is what you use when nothing is printed and no manual exists, not what you use instead of reading the part.

The second rule, from the frame side: Cannondale’s owner’s manual warns that a carbon seatpost can be damaged by clamping force and to use the minimum force necessary. On carbon, the published maximum is a ceiling to approach carefully — not a target to hit.

The four failures torque prevents

• The slipping seatpost.Under-tightened, and every bump drops it a millimeter. The fix is usually not more torque — it is carbon assembly paste and the published figure.
• The creaking bike. Creaks are micro-movement between two surfaces. Bolts at the wrong preload are the most common source, and the noise migrates so convincingly that people replace bottom brackets to fix a stem bolt.
• The stripped thread. Aluminum threads in a steel-bolted world are the soft party. Once stripped, the fix is a helicoil or a new component.
• The cracked carbon.Point loading from an over-tightened clamp is the classic way a carbon bar, post or steerer fails — often days later, under load.

Technique: the part nobody reads

A torque wrench is only as good as how it is used, and Park Tool publishes the rules in its torque wrench use and care guide:

• Wind click wrenches down after use.Park Tool: click-type wrenches “should be wound down to the lowest value on the scale (not zero) when not in use”. Storing a wrench at 20 N·m leaves its spring compressed and its calibration drifting.
• Do not loosen bolts with it.“It is not advisable to use click-type torque wrenches as common ratchet handles to remove fasteners or components.” Use a normal wrench to undo, the torque wrench only to do up.
• Recalibrate.Park Tool suggests every 5000 clicks, or at least once a year for a wrench in frequent use. A home wrench used a few times a month will not hit 5000 clicks in years — but it will still age.
• A damaged thread cannot be torqued.“A cross-threaded bolt will not properly tighten even with a torque wrench.” If it feels wrong going in, stop; the click you eventually get will be friction, not preload.
• Approach slowly and stop at the click.One smooth pull, not a series of jerks, and nothing after the click — a second pull adds torque the tool no longer measures.

The range problem: what home wrenches do not reach

Look at the table below and note where the big numbers are. Stem, seatpost, derailleur and caliper bolts sit between about 4 and 12 N·m — comfortably inside every tool in this guide. Then the drivetrain: Park Tool’s reference table lists a Shimano cassette lockring at 29.4–49 N·m, SRAM at 40, crank bolts at 34–44 N·m (Shimano) and pedals into cranks at 35 N·m minimum (Shimano) or 40 (Campagnolo).

None of the three tools above reaches those figures. That is not a criticism of them — it is the structure of the category, and it is the same ceiling our home workshop tools roundupdocuments in detail. If your job list is mostly cassettes and bottom brackets, you need a bigger, separate wrench, and no amount of care with a 20 N·m tool substitutes for it.

Where torque figures come from, and why brands differ

Notice in the table that two brands rarely agree. Shimano publishes 19.6–29.4 N·m for a stem handlebar binder; Control Tech publishes 13.6–16.3. Deda publishes 8 N·m for a threadless stem steerer bolt; Thomson 5.4; Race Face 6.2. These are not different opinions about the same bolt — they are different bolts, in different materials, with different thread pitches and clamp geometries.

That is the whole reason the “printed on the part wins” rule exists. A table is a starting point when you have nothing else; the component maker’s number is the answer.

What we could not tell you

We have not verified any manufacturer’s torque figure with a calibrated transducer, and no publisher without a test lab has. What this page does is collect published figures, attribute each to the brand that published it, and reproduce the caveats those brands attach — including Park Tool’s own reminder that its table is reference only and that some companies do not publish figures at all.