What is percent error?
Percent error expresses the gap between what you measured and what the value should be, as a share of the correct value. It is a unitless way to say how accurate a measurement was: 2% is a close result, 40% suggests a flawed method or a mistaken reading. Scientists often call the correct value the accepted, theoretical or true value, and the measured one the observed or experimental value.
Percent error formula
Some courses drop the absolute-value bars so the sign shows the direction of the error. A negative result means the measurement came in below the true value, and a positive result means it came in above. This calculator shows both: the signed result first, then its absolute value.
Worked examples
Absolute percent error
A lab measurement gives 56.891 but the accepted value is 62.327. The absolute difference is 5.436. Dividing by 62.327 gives 0.08722, so the percent error is about 8.7217% (signed: -8.7217%, since the reading is low).
Signed percent error
You measure 7 where the true value is 9. The signed error is (7 - 9) / 9 = -0.2222, which is -22.2222%. The negative sign tells you the measurement was too low.
Percent error versus similar measures
| Measure | Base of the comparison | When to use it |
|---|---|---|
| Percent error | The true value | A trusted reference value exists |
| Percent difference | The average of two values | Neither value is more correct |
| Percent change | The old value | Comparing before and after |
Why the true value cannot be zero
The formula divides by the true value, so a true value of zero makes the result undefined. In that situation report the absolute error in the units you measured instead.
Reducing percent error
- Calibrate instruments before use and check them against a known standard.
- Repeat the measurement several times and use the average.
- Use an instrument whose resolution is finer than the quantity you care about.
- Control conditions such as temperature, which can shift readings.
- Record readings at eye level or with a digital readout to avoid parallax.
Interpreting the result
There is no universal cutoff for an acceptable percent error. Introductory physics labs often accept 5% to 10%, while precision machining or analytical chemistry may require well under 1%. Judge your result against the tolerance the task demands and the known limitations of your equipment.
Step by step method
Whenever you compute percent error, follow the same four steps. They keep the arithmetic consistent and make it easy to check your work.
- Subtract the true value from the observed value to get the error.
- Take the absolute value if you only need the size of the error.
- Divide by the true value, not the observed one.
- Multiply by 100 and attach a percent sign.
More worked examples
A student finds the density of an aluminum block to be 2.62 g/cm3, while the accepted value is 2.70 g/cm3. The error is 2.62 - 2.70 = -0.08, and dividing by 2.70 gives -0.02963. The percent error is 2.963%, or -2.963% when signed. This is a good result for a classroom lab.
A kitchen thermometer reads 98.6 degrees C in boiling water at sea level, where the true value is 100 degrees C. The error is -1.4, and 1.4 / 100 = 0.014, so the percent error is 1.4%. A reading that low suggests the thermometer needs calibration.
Absolute error, relative error and percent error
These three terms are related. Absolute error is the raw difference between the observed and true values and keeps the units of the measurement. Relative error divides that difference by the true value, so it has no units. Percent error is the relative error multiplied by 100. In the density example, the absolute error is 0.08 g/cm3, the relative error is 0.02963, and the percent error is 2.963%.
Common mistakes
- Dividing by the observed value instead of the true value, which changes the answer.
- Forgetting to convert both values to the same unit before subtracting.
- Rounding too early. Keep extra digits during the calculation and round at the end.
- Using percent error when no accepted value exists. In that case compare the measurements with percent difference or report the spread of the data.
Rounding and significant figures
A percent error usually has no more significant figures than the least precise measurement used to compute it. Reporting 8.7217% from a ruler that reads to the nearest millimeter suggests a precision the data do not have, so 8.7% is more honest. Check what your course or lab requires, since some instructors want two or three significant figures and others a fixed number of decimal places.