A percentage difference calculator answers a narrower question than most people expect. It compares two values that stand on equal footing — two lab measurements, two suppliers' quotes, two branches' figures — and expresses the gap between them as a percentage of their average. Because it divides by the average rather than by either individual value, it gives the same answer whichever number you write first. That symmetry is the entire point, and it is what separates this measure from percentage change.
Arb Digital built this page because the two are constantly swapped for each other, and the swap is not harmless. Comparing 40 and 60 gives a percentage difference of 40%. Going from 40 up to 60 is a percentage change of +50%. Going from 60 down to 40 is a percentage change of −33.33%. Three different numbers, all correct, all describing the same pair. Which one you should quote depends entirely on whether your two values are peers or a before-and-after.
Percentage Difference Is Not Percentage Change
This is the distinction the page exists for, so it comes first rather than buried below. Percentage change is directional. It has a starting value and an ending value, it divides by the starting value, and it carries a sign that tells you whether things went up or down. Reverse the two and you get a different number, which is correct behaviour: a rise from 40 to 60 is genuinely a bigger proportional move than the fall from 60 back to 40, because the base you are measuring against is smaller.
Percentage difference is symmetric. Neither value is the base. It divides by the mean of the two, it is conventionally reported as a positive number, and swapping the inputs changes nothing. It answers "how far apart are these two things, proportionally?" and it deliberately refuses to say which one is bigger.
The practical test is one question: does one of my two numbers come first in time, or is one of them the reference? If yes, you want our percentage change calculator — revenue this month against last month, a price before and after a rise, a population in two census years. If no, and the two values are simply two readings of the same kind of thing, percentage difference is the honest measure. And if one value is a known true value and the other is an estimate, neither applies: use the percent error calculator, which divides by the true value.
What This Calculator Does
Enter two values and the tool returns the percentage difference as its headline number, alongside the absolute difference and the mean it divided by. It then shows both percentage changes — first to second and second to first — so you can see all three figures side by side and pick the one your situation actually calls for. The breakdown rows add the difference relative to each individual value, so nothing is hidden.
It sits alongside our general percentage calculator, which handles "what is X% of Y" and its relatives, and the average percentage calculator, which deals with the separate problem of averaging several percentages that have different underlying group sizes.
How to Use It
- Enter both values. Order does not matter for the headline result, which is the whole idea. Order does matter for the two change figures in the grid.
- Read the headline as a spread, not a direction. "These two values differ by 40%" is the correct sentence. "The second is 40% higher" is not — that is a different calculation.
- Check the mean shown beside it. Knowing what the percentage is a percentage of is what stops the number being misread.
- Compare with the two change figures. If they are far apart from each other, your two values are far apart, and quoting the wrong measure will mislead by a wide margin.
- Press swap to prove the symmetry. The headline stays put, the two change values trade places. That single demonstration is worth more than any explanation.
The Formula: How It Is Calculated
The formula is percentage difference = |a − b| ÷ ((a + b) ÷ 2) × 100. Take the absolute difference between the two values, divide by their arithmetic mean, and express as a percentage.
Work the default example. With a = 40 and b = 60, the absolute difference is 20 and the mean is 50, so the percentage difference is 20 ÷ 50 = 0.4, or 40%. Swap them and nothing changes, because both |a − b| and (a + b) ÷ 2 are unaffected by order.
Compare that with the change formula, percentage change = (new − old) ÷ old × 100. From 40 to 60 that is 20 ÷ 40 = 50%. From 60 to 40 it is −20 ÷ 60 = −33.33%. The percentage difference of 40% sits between the two, which is not a coincidence — dividing by the mean is exactly the compromise between dividing by the smaller value and dividing by the larger one.
Why Divide by the Mean Rather Than Either Value
Any measure of relative difference needs a denominator, and the choice of denominator is a choice about what you are claiming. Dividing by a says "as a fraction of a". Dividing by b says "as a fraction of b". Both are legitimate, and both build an asymmetry into a comparison between equals — the answer depends on which value you happened to write down first, which for two peer measurements is arbitrary.
Dividing by the mean removes that arbitrariness at a small cost: the result is no longer a fraction of anything you actually measured. It is a fraction of a hypothetical midpoint. Metrology handles this the same way when comparing two instruments with no reference standard between them, and the general principle of stating what a relative quantity is relative to is set out in NIST's guidance on the uncertainty of measurement results.
There is a second, quieter reason. Because it is symmetric, percentage difference is bounded in a way percentage change is not. Percentage change is unbounded above — a rise from 1 to 1,000 is +99,900% — but can never go below −100%. Percentage difference between two positive numbers can never exceed 200%, which it reaches only when one of them is zero. That bounded range makes it far more usable for automated flagging: "alert if two systems' figures differ by more than 5%" behaves the same in both directions, whereas the same rule written with percentage change fires asymmetrically.
The Cases Where It Breaks Down
Percentage difference is undefined when the mean is zero, which happens whenever a = −b. Comparing +5 and −5 gives an absolute difference of 10 divided by a mean of 0, and the calculation has no answer. This tool says so rather than returning infinity or a blank.
More generally, the measure is unreliable whenever the two values sit on opposite sides of zero, even when the mean is not exactly zero. Comparing +10 and −8 gives a mean of 1 and a difference of 18, producing 1,800% — an arithmetically correct figure that describes nothing anyone wants to know. If your quantities can be negative, report the absolute difference in its own units and leave the percentage alone.
The zero case is more common and more interesting. If one value is 0 and the other is any positive number, the percentage difference is always exactly 200%, whether the second value is 3 or three million. That is a genuine feature of the definition rather than a bug — the mean is half the non-zero value, and the difference is the whole of it — but it means the measure carries no information at all when one side is zero.
Percentage Points Are a Third Thing Entirely
If your two values are themselves percentages, none of the above applies cleanly, and this is where reporting goes wrong most publicly. A conversion rate moving from 2% to 3% has risen by 1 percentage point. Its percentage change is +50%. Its percentage difference is 40%. All three sentences are true and all three describe the same event, which is why the phrase "up 50%" without qualification is close to meaningless when the underlying quantity is already a rate.
The convention that resolves it is simple and worth enforcing in your own writing: use percentage points for the arithmetic gap between two percentages, and reserve percent for a proportional comparison. An interest rate rising from 4% to 5% has gone up one percentage point, or by 25% in relative terms. Both are correct; only one of them is what a reader assumes when they see "up 25%". Our ratio calculator is often a clearer way to present these comparisons, because a ratio of 3:2 leaves no room for the ambiguity at all.
Choosing the Right Measure in Practice
Two laboratories measure the same sample and get 4.85 and 5.02. These are peers with no reference value, so percentage difference is right: the gap is 0.17 over a mean of 4.935, which is 3.44%. Reporting "lab B is 3.5% higher" would imply lab A is the reference, which it is not.
A supplier quotes 4,200 and a competitor quotes 4,700. Again peers, again percentage difference: 500 over a mean of 4,450, or 11.24%. But once you have chosen one supplier and are evaluating a switch, the incumbent becomes the base and percentage change is the right measure instead.
A dataset's mean was 120 last quarter and 138 this quarter. That is a before-and-after, so percentage change applies: +15%. Using percentage difference here would produce 13.95% and quietly discard the direction, which is the single most important part of the finding. When you are comparing distributions rather than single figures, the descriptive statistics calculator and the standard deviation calculator give a fuller picture than any single percentage can, and the NIST/SEMATECH e-Handbook of Statistical Methods covers the formal comparison tests.
Arb Digital publishes hundreds of free calculators covering percentages, statistics, algebra and finance — all free to use, with no sign-up and no limit on how often you run them.
Browse All Free Tools Suggest a ToolCommon Mistakes to Avoid
- Using it for a before-and-after. If one value came first in time, percentage change is the measure, and its sign is information you should not throw away.
- Using it when one value is a reference. A known correct value belongs in the denominator on its own. That is percent error, not percentage difference.
- Attaching a direction to it. The result is symmetric by construction, so "40% higher" is not a sentence it can support.
- Applying it to values of opposite sign. The mean can approach or reach zero and the result becomes arbitrarily large or undefined.
- Confusing it with percentage points. When both values are already percentages, state the gap in percentage points and the relative move separately.
Related Free Tools From Arb Digital
Use the percentage change calculator for directional before-and-after comparisons, the percent error calculator when one value is a known reference, the general percentage calculator for everyday percentage arithmetic, and the average percentage calculator when several percentages have to be combined correctly. More live in the free tools hub.
Frequently Asked Questions
Take the absolute difference between the two values, divide by their arithmetic mean, and multiply by one hundred. For forty and sixty that is twenty divided by fifty, which gives forty percent. Swapping the two values leaves the answer unchanged.
Percentage change is directional and divides by the starting value, so it carries a sign and gives different answers in each direction. Percentage difference is symmetric and divides by the mean of the two values, so it gives one answer regardless of order and states no direction.
Use it when the two values are peers with no reference or starting point between them, such as two laboratory readings of the same sample or two competing quotes. Use percentage change whenever one value comes first in time or serves as the base for the comparison.
Because dividing by either individual value would make the result depend on which one you wrote first, and for two equal-status values that choice is arbitrary. Dividing by their mean removes the asymmetry, at the cost that the denominator is a midpoint rather than something you measured.
Yes. For two positive values it can reach up to two hundred percent, which happens only when one of them is zero. Values of opposite sign can produce far larger figures, which is a sign the measure is not appropriate for that data.
The percentage difference is exactly two hundred percent, whatever the other value is, because the mean is half the non-zero value and the difference is all of it. The measure therefore carries no useful information when one side is zero.
Not by convention. The formula takes the absolute value of the gap, so the result is reported as a positive number. If you need to say which value is larger, you need percentage change or a plain statement of the two figures.
Percentage points measure the arithmetic gap between two percentages. Moving from two percent to three percent is a rise of one percentage point, a percentage change of fifty percent, and a percentage difference of forty percent. Naming which one you mean prevents the most common reporting error there is.
This calculator is provided for study and for checking your own working. It describes how each measure is computed and does not recommend which figure to publish in any particular report.
