How to calculate percentage error

Subtract the accepted value from the experimental value and drop the sign. Divide by the accepted value, then multiply by 100.

percentage error = |experimental - accepted| / |accepted| × 100

  1. Find the difference: experimental minus accepted.
  2. Take the absolute value, so the gap is positive.
  3. Divide by the accepted value.
  4. Multiply by 100.
  5. Example: |9.8 - 10| / 10 × 100 = 0.2 / 10 × 100 = 2%.

The accepted value cannot be zero, because the formula divides by it.

Worked examples of percent error

Each row uses the same formula. The error is the gap between your measurement and the reference, as a share of the reference.

ExperimentalAcceptedPercentage error
9.8102%
48504%
1051005%
98.51001.5%
2.652.701.85%

Experimental value vs accepted value

The experimental value is the number you measured or calculated. The accepted value is the reference number: a known constant, a theoretical result, or a label value. Some textbooks call it the true or theoretical value.

Order matters because the accepted value is the divisor. For 105 against an accepted 100, the error is 5 / 100 = 5%. Swap them and you get 5 / 105 = 4.76%, which is wrong for this question.

Signed error and absolute error

This calculator reports the absolute percentage error, so an overestimate and an underestimate of the same size give the same number. Some classes keep the sign to show direction.

A signed error for 9.8 against 10 is (9.8 - 10) / 10 × 100 = -2%. A negative sign means the measurement came in below the accepted value. Percentage error can also exceed 100%. A reading of 30 against an accepted 10 is a 200% error.

Percentage error, difference, and change

These three measures use similar steps but divide by different bases. Pick the one that matches your data.

MeasureDivides byUse it when
Percentage errorThe accepted valueYou compare a measurement with a known standard
Percentage differenceThe average of both valuesNeither value is the standard
Percentage changeThe starting valueA value moved from old to new over time

Open percentage difference

Not the same as

Percentage error needs an accepted value. The other measures do not.

Percentage-error questions

Subtract the accepted value from the experimental value, take the absolute value, divide by the accepted value, and multiply by 100.

2%. The gap is 0.2, and 0.2 / 10 × 100 = 2.

Yes. The absolute difference removes the sign, so overestimates and underestimates of the same size give the same percent.

Not with this calculator, which uses the absolute value. A signed version, (experimental - accepted) / accepted × 100, gives a negative number when you measure below the accepted value.

Because one value is treated as correct. Dividing by the average of two values is percentage difference.

Open percentage difference

It depends on the method and the field. Your instructor, lab manual, or tolerance spec sets the limit. A smaller percent error means a closer match to the accepted value.

The formula cannot run. You cannot divide by zero, so percent error has no meaning for a zero reference.

What-percent is part over whole. Percentage error is a gap over an accepted value.

Open what-percent