How to Choose and Use a Torque Wrench

A torque wrench doesn't measure how tight a bolt is. It measures how hard you twist. What actually holds a wheel or a head on is bolt tension, and most of your effort never gets there. That is why the number you set has to come from the car's own manual, why the thread condition matters, and why the way you use and store the wrench decides whether its reading means anything.

Where your effort goes, by Norbar's figures: most of it is friction.

What a torque wrench actually controls

Norbar, the torque-tool maker, puts it plainly: bolt tension, not torque, is the critical factor, and torque is "an indirect means of establishing tension".

Most of the twist goes into friction. Norbar's figures:

• As much as 50% of the torque you apply is spent overcoming friction under the bolt head.
• Another 30% to 40% is lost to friction in the threads.
• As little as 10% does the useful work of stretching the bolt.

Those are upper and lower bounds, not a fixed split. The point is that small changes in friction make big changes in tension. Norbar also notes that removing the oil film new fasteners usually carry changes the tension you get at a given torque. Dry or oiled threads give a different result at the same setting, which is why the manual tells you which one it means.

The types, and how each one measures

The standard for hand torque tools, ISO 6789-1:2017, sorts them into two families:

• Indicating (Type I): the wrench shows you the torque "by means of a mechanical scale, dial or electronic display". Beam wrenches, dial wrenches and most digital wrenches are here.
• Setting (Type II): you set a value and the wrench signals when you reach it, with an impulse that briefly drops the applied torque. Click and slipper wrenches are here.

What you'll actually see in a shop:

• Click-type: a spring holds a detent until the set value, then it releases. Gedore describes it as a click you hear, an impulse you feel and about 3° of movement. The most common choice for home garages.
• Beam: a handle that bends under load while a separate pointer stays straight over a scale. Simple, but you have to read it while you pull.
• Slipper: the handle slips free at the set torque, then resets, so it can't be over-pulled. Gedore's TSN 55 is pre-set with no scale and tightens clockwise only.
• Digital: shows the value as you tighten and alerts you at the target. Some, like Norbar's ProTronic, also measure angle, which you need for torque-plus-angle bolts (see Torque-plus-angle and "Replace").

For most home mechanics a click-type wrench is the practical choice.

How accurate is "accurate"?

"±4%" is not a universal number. ISO 6789-1:2017 allows:

• ±4% for click-type wrenches (Type II classes A, B and C) whose maximum is above 10 N·m.
• ±6% for the same click types with a maximum of 10 N·m or less.
• ±6% for beam (flexion-bar) wrenches.

Direction matters too. Bahco states ±4% tightening and ±6% in reverse for its 7455 click wrenches, and Snap-on gives ±4% clockwise and ±6% counter-clockwise for its QD wrenches. Some makers claim better: Proxxon and Stahlwille state ±3% for some models, and Norbar's ProTronic electronic wrench claims ±2% from 20% to 100% of full scale.

Where on the scale to work

Under ISO 6789-1:2017, a graduated click or beam wrench is specified "From the lowest marked value to 100 %" of its maximum. The range from zero to the lowest mark has to be marked as outside the specified range. Norbar says that lowest mark is "usually 20% of maximum".

The US standard still covers the range below 20%, with a much looser tolerance. Snap-on's brochure explains that it allows 4% of reading between 20% and 100% of capacity, and below 20% an error equal to 0.8% of the wrench's capacity. Its own example: at 5% of full scale, the reading "may result in up to 16% error".

The practical rule, from Snap-on: choose a wrench so your working range "falls near the mid point of its capacity". Gedore adds that the working value "should not be permanently at the limit of the measuring range". If you work on both small engine bolts and wheels, that usually means two wrenches, not one with a huge range.

What to look for when buying

• Stated accuracy and ISO 6789 conformance. Anything without a stated accuracy is a guess.
• A declaration of conformance, not necessarily a certificate. Since the 2017 standard, makers supply a declaration of conformance with new wrenches instead of a calibration certificate, Norbar says. ELORA calls the declaration "quite sufficient for most users". A calibration certificate is a separate service that measures your individual wrench.
• The right range for the jobs you do most (see above).
• Direction. Many torque wrenches only read clockwise, Norbar warns. If you'll ever torque a left-hand thread, check the wrench's specification first.
• Units you can read. Manuals may give N·m, lbf·ft or kgf·m. A dual scale saves conversions.

How to use a torque wrench correctly

• Get the value from the car's manual. Snap-on's own instructions tell you to take torque values from the equipment manufacturer.
• Run the fastener down first with a normal ratchet. Tesla's Model X service manual allows an impact or electric wrench "only during the initial threading" of wheel nuts, then a torque wrench for the final value.
• Hold it where it's marked. Norbar says to centre your hand over the marked load point. Snap-on says "keep hand centered on the handle". Don't choke up the shaft.
• Pull smoothly. Snap-on asks for a "slow steady force". Stahlwille says to pull without interruption, especially in the final phase.
• One click is enough. Norbar's rule: extra clicks apply extra torque. Gedore says that pushing on after the roughly 3° release movement increases the torque. At the click, "Stop pulling and allow the wrench to reset" (Snap-on).
• Follow the pattern. For wheels, Tesla says to tighten "evenly using a diagonal sequence" and Volvo says to tighten the bolts crosswise.
• Exercise a click wrench that has been sitting. Snap-on tells you to set its click wrenches to 50% of full scale and exercise them at least three times in both directions before use.

Grip at the marked point, pull smoothly, stop at the first click.

Extensions and crow's feet change the torque

The scale setting is the torque at the wrench's square drive, Snap-on explains. Anything that moves the fastener further from the handle changes what the bolt actually gets.

Norbar's formula for an extension fitted at the head, which puts the nut further out than the square drive: setting = torque wanted × normal length ÷ extended length. Worked through: if you need 130 N·m at the nut, the wrench's normal length is 500 and the extension makes it 650 (in the same unit), set the wrench to 130 × 500 ÷ 650 = 100.

• Crow's-foot fittings also change the torque, and Norbar says "a suitable offset is applied to the set torque".
• For universal joints, Norbar's ProTronic manual limits the offset to 15° from perpendicular. According to one retailer's instruction sheet, universal joints are best avoided because the error is hard to determine, so avoid them where you can.

A pipe on the handle is a different problem. It doesn't move the fastener, so it doesn't call for a lower setting and the formula above doesn't apply: you just pull with less force. Norbar still says never to put a pipe or any other kind of extension on the handle, because doing so "can seriously damage the tool and make it inaccurate", and it's a safety hazard. It also takes your hand off the marked load point (see Hold it where it's marked).

It's not a breaker bar

Snap-on: "Do not use a torque wrench to break fasteners loose." Gedore says a torque wrench is "generally intended for controlled tightening" and to use a loosening tool on seized bolts.

Some makers allow controlled loosening within the wrench's rating: Stahlwille's FAQ permits it if the load stays within the nominal value, and Norbar says to undo bolts only within the wrench's maximum and "If in any doubt, use another tool for loosening bolts." A rusted wheel nut is exactly that doubt. Stahlwille warns that loads more than 25% above the maximum can make readings inaccurate without you noticing.

Units: convert with the exact factor

• One pound-force foot (lbf·ft) equals 1.355 818 N·m (NIST).
• Snap-on's rounded workshop factors: N·m × 0.7376 = lbf·ft, and lbf·ft × 1.356 = N·m.

Care and calibration

Storage: lowest marked setting, never zero

For spring-type click wrenches, Norbar says to "wind it down to the minimum scale setting and never to zero", and that winding to zero can affect calibration. Snap-on says to store it "at its lowest scale setting" and "Do not force handle beyond lowest setting." Gedore says returning to the minimum "relieves the mechanism during storage".

How much it matters is less clear than forum lore suggests. In a small Norbar test, four wrenches stored for a month at maximum and minimum all stayed in tolerance, and Norbar says the sample is too small to be conclusive. Some designs, like Stahlwille's bending-beam MANOSKOP 730N, need no reset at all. Check your wrench's manual. The safe default for a spring click wrench: lowest marked value, never below it.

Drops and overloads

• Norbar's ProTronic manual: "Verify calibration if dropped."
• Norbar says a wrench overloaded by 25% or more above its maximum should be recalibrated.
• ISO 6789-2 requires calibration after an overload, after repair, or after any improper handling that might affect performance.

If it has been dropped, used to break a seized bolt, or leaned on past its maximum, get it checked before trusting it on wheels or engine work.

How often to calibrate

ISO 6789-2:2017 says that if you have no control procedure of your own, "12 months, or 5,000 cycles, whichever occurs first" may be taken as the default calibration interval, and "The interval starts with the first use of the torque tool." It is a default, not a law: the standard allows shorter intervals, and makers differ. Gedore recommends "at least once a year or after 5,000 load cycles". Snap-on's UK brochure says "Recalibrate every 6 months or more depending on use."

The torque value belongs to the car

The same job has different numbers on different cars. Wheel nuts and bolts, from the makers' own manuals:

Kia's 2027 US owner's manual gives 79–94 lbf·ft (11–13 kgf·m). There is no single "typical" wheel torque to memorise: look up yours.

Dry or oiled threads

• Mazda MX-5: "Applying oil or grease to wheel nuts and bolts is dangerous", because the nuts could loosen while driving and the tyre could come off.
• Volvo V90: "Do not use lubricant on the threads of the wheel bolts." Polestar and Kia say the same.

Oil changes the tension you get at the same setting, so follow your manual, not a rule of thumb.

Re-check after a wheel change

• Ford: retighten to the specified torque within 100 mi (160 km) after any wheel disturbance.
• Volvo: the bolts "may need to be re-tightened several days after the change".
• Ford also warns that dirt or corrosion on the wheel and hub mounting faces can let the nuts loosen, so clean them before fitting.

Torque-plus-angle and "Replace"

Some specs don't end at a torque value. Tesla's Model Y service manual tightens the rear drive unit LH mount bolts to 35 N·m and then 55 degrees, with a torque wrench that measures angle, and marks the bolts "Replace". Tesla's torque tables carry a Reuse/Replace column for every fastener.

Those bolts are stretched on purpose. ElringKlinger, the gasket maker, says torque-plus-angle cylinder-head bolts go beyond the yield point, and "On no account may used bolts with plastic elongation be re-used." If the spec says angle and Replace, buy new bolts.

Drain plugs too

Engine oil drain plug torque runs from 8 N·m to 50 N·m across primary OEM documents, depending on the engine and the plug. See How to Change Your Oil.

Related guides

• Essential Tools for Your Home Garage Workshop
• How to Change Your Oil
• How to Change Spark Plugs
• How to Change Brake Pads: Step-by-Step Guide
• How to Rotate Your Tires at Home