How to Choose a Torque Wrench: Range, Type, and What to Skip

A torque wrench is one of the few tools where buying the wrong size is worse than not owning one, because a wrench used outside its accurate range gives you a number you will trust and should not.

Pick the range first

Accuracy is specified as a percentage of reading, and it degrades badly in the bottom 20% of a wrench’s range. So the range matters more than the brand.

Work	Useful range	Drive
Spark plugs, valve covers, brake calipers	10–80 ft-lb	3/8”
Lug nuts, suspension, crank bolts	30–250 ft-lb	1/2”
Small engine, electronics, bike components	2–20 Nm	1/4”

If you buy one wrench, buy the 3/8-inch in roughly 10–80 ft-lb. It covers most routine maintenance. Lug nuts at 100 ft-lb need the 1/2-inch — do not stretch the smaller wrench to reach them.

Types, and what they are worth

Click (micrometer) wrenches are the default: set the handle, pull until it clicks. Good ones hold ±4% for years. This is what you want.

Beam wrenches are cheap and never drift out of calibration because there is nothing to drift. The catch is you must read a scale while pulling, which is awkward in a wheel well.

Digital wrenches add an angle mode, which matters if you do torque-plus-angle head bolts. Otherwise you are paying for a battery.

Three habits that keep it accurate

1. Wind it back to the low end of its range — not to zero — before storing. Leaving the spring compressed is what kills click wrenches.
2. Pull smoothly and stop at the first click. A second pull adds torque you cannot measure.
3. Never use it to loosen anything. Breaking a seized bolt loose through a torque wrench is how the mechanism gets damaged.

Calibration

Send it out roughly every 5,000 cycles or every couple of years for anything where a failure is dangerous — wheels, suspension, engine internals. For valve covers, honestly, the drift will not hurt you.

Reading the accuracy spec properly

Torque wrenches are specified as a percentage of the reading, not of the full scale. A wrench rated ±4% set to 80 ft-lb is accurate to about 3 ft-lb; the same wrench set to 15 ft-lb is accurate to 0.6 ft-lb — in theory.

In practice, accuracy degrades sharply in the bottom fifth of the range, which is why the range table above matters more than the brand name. A 20–150 ft-lb wrench set to 25 ft-lb is operating where its calibration is least trustworthy, even if the scale goes that low.

Two consequences worth internalising:

• Do not buy one wrench to cover everything. A 10–150 ft-lb wrench looks efficient and is unreliable at both ends.
• Buy the range that centres on your common jobs. If most of your work is 30–60 ft-lb, a 10–80 wrench puts that squarely in its accurate middle.

Nm or ft-lb

Buy whichever matches the specs you work from, and avoid converting under a car. 1 ft-lb is 1.356 Nm. Most wrenches show both scales, but the primary scale — the one with the finer graduations — is the one to set from. If your service data is metric and your wrench is primarily imperial, you will make an arithmetic error eventually.

Technique that changes the reading

The number on the handle assumes you use it a particular way:

• Pull at the centre of the handle. Gripping further out increases leverage and overtightens; the calibration assumes a specific effective length.
• Pull perpendicular to the wrench, in one smooth motion. Jerking it overshoots before the mechanism releases.
• Do not use an extension bar on the handle. It changes the effective length and invalidates the setting entirely.
• A crow-foot or offset adapter changes the reading if it extends the length. At 90 degrees to the wrench axis it does not; in line with it, you need to recalculate. If you are unsure, do not use one for a critical fastener.
• Threads must be clean and, where specified, lubricated. Torque is a proxy for bolt stretch, and a dry or corroded thread can absorb most of your applied torque in friction, leaving the fastener loose at the correct reading.

That last point is the one people find surprising. Torque specs assume a stated thread condition. A rusty bolt tightened to spec is not tightened to spec.

Torque-to-yield fasteners

Some bolts — most head bolts, many suspension and crank fasteners on modern engines — are designed to be tightened past their elastic limit. Their spec is a torque figure followed by an angle: “40 Nm plus 90 degrees”.

These are single-use. Reusing a torque-to-yield bolt is how a head gasket job fails six months later. Check the service data before reusing anything critical, and buy the bolt kit when it calls for one. An angle gauge or a digital wrench with an angle mode is required for these; a click wrench alone cannot do it.

What to spend

A reputable 3/8-inch click wrench in the 10–80 ft-lb range costs modestly and will hold calibration for years with reasonable care. The very cheapest wrenches are frequently out of spec from new, which is worse than useless — it gives you a number you will trust. If the budget is tight, a beam wrench is the honest choice: less convenient, but it cannot drift.

Storing and checking it

Wind it back to the low end of its range — not to zero, which on many designs unloads the spring incorrectly — and keep it in its case, away from damp and impact. A torque wrench dropped onto concrete should be treated as out of calibration until proven otherwise.

You can sanity-check one at home with a beam and a known weight: apply a measured force at a measured distance from the drive and compare. It will not certify the tool, but it catches gross error, which is the failure that matters.