Percent Error Calculator
This percent error calculator compares an observed value with an accepted nonzero reference. It reports absolute percent error, signed relative error using the reference magnitude, and the original-unit difference between the two values.
By Dr. Hassan Khames EliwaLast updated
Result
- Signed relative error
- -2.000000%
- Observed minus reference
- -2
How it works
Anchor the comparison to a reference
Absolute percent error = |observed - reference| / |reference| × 100. The numerator measures the size of the discrepancy, while the denominator expresses that discrepancy relative to the accepted benchmark. Because both quantities use matching units, the ratio is dimensionless and can be written as a percentage.
The reference is not simply whichever of the two numbers happens to be larger. It should be the accepted value, theoretical prediction, calibrated standard, or other explicitly chosen benchmark. Reversing the fields changes the denominator and usually changes the answer, even though the absolute difference between the two numbers stays the same.
Keep direction visible
Absolute percent error is never negative. To preserve direction, the calculator also reports signed relative error = (observed - reference) / |reference| × 100. Under this stated convention, positive means the observation is numerically above the reference, and negative means it is below. Using the reference magnitude keeps that direction rule consistent even for negative references.
Some disciplines divide signed relative error by the signed reference instead. Those conventions agree for positive references but differ for negative ones. Label the formula when sharing results so a reader can reconcile the sign without guessing which definition was used. The raw Observed minus reference output provides an additional check.
Respect the zero-reference limitation
A zero reference cannot be used because division by zero is undefined. References with magnitude below 1e-12 are also outside this tool's supported range. Both inputs must be finite and have magnitude no greater than 1e12. These explicit bounds prevent unstable or unrepresentable comparisons from being displayed as ordinary answers.
Percent outputs normally show six decimal places, while the raw difference shows up to ten. Very small nonzero results and very large results switch to scientific notation. A rounded display of a small discrepancy should not be treated as evidence of perfect measurement agreement.
Worked example
Example 1: compare a measured mass
A scale reports 98 grams for a reference mass accepted as 100 grams. Enter observed 98 and reference 100. Observed minus reference is -2. Signed relative error is -2.000000%, and Absolute percent error is 2.000000%. The measurement is two grams below the reference, with a discrepancy equal to two percent of the reference mass.
Example 2: check a slightly high reading
An instrument reports 51 units against an accepted reference of 50 units. Enter observed 51 and reference 50. Observed minus reference is 1. Signed relative error is 2.000000%, and Absolute percent error is 2.000000%. The same absolute percentage as the first example now describes an above-reference reading, which is why the signed output is useful.
Expert tips
Common mistakes
- Confusing percent error with percent difference. Percent difference commonly uses a symmetric denominator such as the average magnitude. Percent error privileges an accepted reference, so the two measures answer different questions.
- Ignoring a small denominator. A modest original-unit discrepancy can produce a huge percentage when the reference is close to zero. Inspect the raw difference and the practical measurement scale before calling the result catastrophic.
- Calling a low error proof of accuracy everywhere. Agreement at one test point does not establish an instrument's performance across its full range. Calibration may require multiple points, repeat measurements, and an uncertainty analysis.
Frequently asked questions
Can percent error exceed one hundred percent?
Yes. If the absolute discrepancy exceeds the reference magnitude, absolute percent error exceeds one hundred percent. The formula has no upper limit of one hundred.
Can the reference be negative?
Yes, within the supported magnitude bounds. Both percent calculations use the magnitude of the reference as the denominator, and the signed result follows observed minus reference.
What happens when the observation equals the reference?
All three results are zero, provided the reference is valid and nonzero. This indicates agreement for the supplied numbers, not necessarily absence of measurement uncertainty.
Why is a zero reference rejected instead of showing infinity?
A percentage relative to zero is undefined under this formula. Reporting the raw difference or choosing a meaningful nonzero scale is more informative than presenting an infinite percentage.
Is absolute percent error the same as uncertainty?
No. Error compares an observation with a reference, while uncertainty characterizes the range of values reasonably associated with a measurement. A single observed discrepancy does not estimate all uncertainty sources.
Related calculators
Related guides
MathPercentages people get wrong: change, difference and pointsThe denominator determines the answer. Learn to distinguish percentage change, symmetric difference and percentage-point movements.14 November 2026 · 7 min read
MathStandard deviation without the jargonThe average tells you the centre. Standard deviation helps describe the spread, provided you know what the data represents.14 November 2026 · 7 min readAbout the author
Dr. Hassan Khames Eliwa holds a PhD in Construction & Built Environment and has more than 20 years of international project controls experience. He is Director of Eliwa Project Controls Limited in Auckland and founder of PMMilestone.org and PMMilestone.com.
Buy me a coffee