The formula, and the one decision in it
Percent error is |measured − accepted| ÷ |accepted| × 100. The only judgement call is which value goes on the bottom, and it is always the accepted one — the known-good reference. Dividing by the measurement instead inflates or deflates the result depending on which way you were wrong, which is exactly the bias the measure exists to avoid.
The conventional form uses absolute value, so percent error is never negative. The signed figure is shown too, since knowing whether you read high or low is often the more useful diagnostic.
Percent error versus percent difference
Use percent error when one value is authoritative — a textbook constant, a calibrated instrument, a known standard. Use percent difference when you have two measurements and neither is privileged; it divides by their average instead, so swapping them does not change the answer.
Lab reports usually want percent error. Comparing two people's readings of the same thing usually wants percent difference.
Frequently asked questions
What counts as a good percent error?
It depends entirely on the field. School experiments often accept under 10%; analytical chemistry may want under 1%; engineering tolerances vary by component. The number only means something against the standard for your context.
Is anything uploaded?
No. Your measured and expected values stay on your machine.
Something wrong with this tool, or an idea for it? Tell us