What Does ±4% Accuracy Mean on a Torque Wrench?

Two readings of the spec go around. One says ±4% is a fixed band, 4 percent of the wrench's maximum, the same at every setting. The other says it is 4 percent of whatever you set. The second is right, and the standard says so in its definitions. But that reading leaves out two limits that sit in the same line of the spec: the direction of the pull, and the part of the scale it covers. Below, a 150 ft-lb wrench is worked through at three settings in both directions, and then the stretch of scale the spec does not cover.

What the accuracy figure measures

A click torque wrench lets you set a torque on the handle. When the pull on the fastener reaches that value, the head breaks over with a click. The accuracy figure is how far the real torque at the click may sit from the number on the scale.

Tekton defines it in plain words: "the maximum expected error between the torque value on the wrench's dial and the actual force the wrench exerts on a fastener."

The standard is stricter about the word. ISO 6789-1:2017 calls the figure the maximum permissible deviation. It defines it as the largest "relative measurement deviation" of the torque measured on the test rig, "measured as a percentage of the target torque value on the torque tool." A note adds that "accuracy" is still used for it, though it is "not technically correct."

Two terms in that definition carry the rest of this page:

• Target torque is the value set on the wrench. The percentage is taken from it.
• Relative deviation means the limit is a share of the setting, not a fixed number of ft-lb.

So a ±4% wrench set to 50 ft-lb may click anywhere from 48 to 52 ft-lb. That is Tekton's own example for its 3/8 in 24330, a 10 to 80 ft-lb wrench. It prints the 50 ft-lb window as 4 ft-lb wide. If the 4 percent came from the 80 ft-lb maximum, the window would be 3.2 ft-lb either way, 6.4 ft-lb wide.

Why the full-scale reading keeps circulating

The spec line invites it. Wright prints its micrometer wrench accuracy as "± 4% clockwise, ± 6% counterclockwise direction @ 20% to 100% of scale range." The word scale comes right after the percentage. Read fast, it sounds as if the 4 percent belongs to the scale.

It does not. The "20% to 100% of scale range" part tells you where the promise holds. The "±4%" part tells you how big it is at the setting. They are two separate limits in one sentence.

The Accredited Labs post on this question gets the main point right: the figure "is calculated relative to the setting/target value, not the full scale of the tool." Its examples hold up too: 96 to 104 ft-lbf at a 100 setting, and a 20 ft-lbf spread at 250. Yet it cites no torque standard. It does not mention the counterclockwise figure. And it says most wrenches do best in the middle 20 to 80 percent of their range. The makers' specs here run to 100 percent, and both standards check the wrench at full scale. Nothing in these documents shuts out the top fifth of the scale.

Worked through on a 150 ft-lb wrench

Take a 1/2 in drive click wrench with a scale from 20 to 150 ft-lb (27 to 203 N·m). Give it Wright's micrometer rating: ±4% clockwise and ±6% counterclockwise, from 20 to 100 percent of scale range.

First, find the covered range. 20 percent of 150 ft-lb is 30 ft-lb. So the spec covers 30 to 150 ft-lb, even though the scale starts at 20.

Now work the window at the three points where a wrench like this is checked: 20, 60 and 100 percent.

Read the table across and three things stand out.

The window grows with the setting. At 30 ft-lb the wrench may be off by 1.2 ft-lb. At 150 it may be off by 6. In percent it is the same at every row. In ft-lb it is five times wider at the top.

The full-scale reading is wrong in a costly place. At 150 the two readings agree, which is why the mix-up survives. At 30 ft-lb the full-scale reading gives 24 to 36 ft-lb. That is ±20 percent of the setting, five times the real spec. Someone who believes it will think a good wrench is useless at low settings. Or they will accept a check result that the real spec fails.

Counterclockwise is half as loose again. ±6% against ±4% means every window in that column is 1.5 times the clockwise one. At 150 ft-lb, 18 ft-lb wide against 12.

Now set the wrench to 20, its lowest mark. The table has no window for that row, and that is the honest entry. The setting is 13.3 percent of maximum, under the 20 percent line. The spec says nothing about it.

The counterclockwise figure

Some click wrenches only measure one way. Tekton's 24330 "drives in both directions and measures torque in the clockwise direction." Turned the other way it is a ratchet, not a torque wrench.

Dual-direction wrenches measure both ways, for left-hand threads. Both makers here rate them looser that way:

Tekton works its 1/4 in wrench at 100 in-lb: an 8 in-lb window clockwise (96 to 104) and 12 in-lb counterclockwise (94 to 106). That is the same arithmetic as the table above, at a smaller size.

Neither maker gives a reason for the wider limit. Both cite ASME B107.300, the torque instrument standard. The counterclockwise figure here is as the makers state it.

ISO 6789-1:2017 takes a different line. For a Class A click wrench (adjustable, with a scale), "the maximum permissible relative deviation shall be met in each direction specified by the manufacturer." The ISO figure does not loosen for counterclockwise.

The ±6% is not a rule for every two-way wrench either. Wright's own dial wrenches, which use a torsion beam and a needle, are "± 4% both ways."

What the spec leaves unstated below 20 percent

On a wrench rated at 20 to 100 percent of scale, the bottom fifth of the scale has no accuracy figure. That is not the same as saying it is bad there. It means the maker has made no promise.

The gap is often printed on the wrench. Here is where the 20 percent line falls on four scales from the documents above:

GearWrench says it outright: its ±3% is "in the upper 80% range of maximum torque." That excludes the bottom 20 percent in the same way.

The Tekton 24330 row is the plainest case. Tekton says its calibration evaluates each wrench "at three set points along its full range": 20, 60 and 100 percent, with three readings at each. The lowest point checked is 16 ft-lb. The scale goes down to 10. The page never says what happens from 10 to 16, so this page does not either.

Two practical points follow.

The lowest mark is a storage mark, not a working one. Tekton's scales print "STORE AT 10 FT.-LB." and "STORE AT 10 IN.-LB." at the bottom. That is where the spring should rest, as covered in whether to store a torque wrench at zero or the lowest setting. It is not where the wrench is rated.

Pick the wrench by where the setting falls. If a fastener calls for 25 ft-lb, the 150 ft-lb wrench is outside its spec there. On the Tekton 24330, 25 ft-lb sits inside the span Tekton checks (16 to 80), with a window of 24 to 26 ft-lb. For small fasteners the jump is bigger. The 36 in-lb J-hook bolts in crossover tool box J-hooks are 3 ft-lb. That is below every ft-lb wrench here. On the TRQ21101 it is 24 percent of maximum, above the 20 percent point, with a window of 34.56 to 37.44 in-lb.

How ISO 6789 draws the line differently

Tekton's 20, 60 and 100 percent points belong to ASME B107.300, as Tekton describes it. The international standard puts the bottom of the range somewhere else.

ISO 6789-1:2017 sets the covered range for a Class A click wrench "from the lowest marked value to 100 % of the maximum torque value." No 20 percent line. If the scale starts at 13 percent, the spec starts at 13 percent.

ISO 6789-2:2017 then checks the wrench at that lowest point. As Atlas Copco summarizes it, the first calibration point for Class A, D and G click wrenches is the tool's minimum capacity, then 60 percent, then 100 percent. It takes five results at each point, against Tekton's three, with a 4 percent limit.

The 10 N·m split in the ISO table matters for small wrenches. ISO 6789-1 allows ±6 percent for a Class A wrench whose maximum is 10 N·m (88.5 in-lb) or less, and ±4 percent above that. The TRQ21101 tops out at 150 in-lb (16.9 N·m), so on the ISO table it would sit in the ±4 percent column.

None of the maker pages here claims ISO 6789. They all name ASME B107.300. So for them, the 20 percent line is the one that applies. A wrench whose certificate cites ISO 6789-1:2017 is covered down to its lowest mark.

Where this answer stops

This page reads the published accuracy figure. It does not tell you that a given wrench meets it today.

The figure describes a calibrated wrench on a test rig. A wrench that has been dropped or used past its maximum needs a calibration check before its number means anything. The intervals the makers set, and what to do after a drop, are covered in the storage guide.

The 4 percent covers the wrench alone, not the joint. Adapters, extensions and the way the handle is pulled sit outside it. None of the sources here puts a number on those, so this page does not either.

Below 20 percent on an ASME-rated wrench, the answer is not known from the spec. Nothing in these documents puts a number on it. Do not fill the gap with 4 percent, or with a guess.

ASME B107.300 itself was not read for this page. The ASME figures above are as Wright, Tekton and GearWrench state them. ISO 6789-1 is read from its published preview pages, which include the definitions, the range clause and the deviation tables.

On a safety-critical fastener, such as wheels, brakes, steering or lifting gear, the vehicle or equipment maker's torque spec and procedure come first. Use a wrench whose covered range includes that value, and whose calibration is in date.

Reading any torque wrench spec in four steps

1. Find the percentage and the direction. ±4% clockwise is not ±4% both ways. If no direction is given, check whether the wrench measures counterclockwise at all.
2. Find the range it covers. Look for "20% to 100%", "upper 80%" or "lowest marked value." If none is given, the spec does not say, and this page cannot tell you.
3. Multiply the setting by the percentage. 60 ft-lb at ±4% is ±2.4 ft-lb, so 57.6 to 62.4. Do not multiply the maximum.
4. Check the setting is inside the range. Divide it by the maximum. Under 0.2 on a 20 to 100 percent spec means a smaller wrench.

The same habit of reading the number against what it measures runs through caliper accuracy vs resolution, Class II on a tape measure and 0.5 mm/m on a spirit level. Each spec states a limit, a basis and a range, and each one is misread when one of the three is dropped.