Grade Curve Calculator
The test turned out harder than planned? Curve it fairly: square root, flat points, or scale-to-highest — with the exact lift each method gives, so you can pick before you commit.
Spotted a mistake or have an idea? — errors get fixed and credited publicly.
Last reviewed: July 20, 2026 · Formulas verified by automated tests before every release.
The three curves, honestly compared
- Square root lifts the bottom most (49 → 70) and the top least (90 → 94.9) — compresses the spread, saves failing grades.
- Flat +N is transparent and easy to defend in front of a class — everyone sees the same bonus.
- Scale to highest re-anchors the test: the best real performance becomes the new 100%.
√-curve: 10 × √raw · Flat: raw + N (max 100) · Scaled: raw ÷ highest × 100
Teacher's checklist before curving
- Was the test the problem (everyone missed Q7?) or the preparation? Curve tests, not effort.
- Pick one method for the whole class — mixing methods per student is indefensible.
- Announce the curve after grading, not before the test — pre-announced curves change how students study.
Frequently asked questions
What is a square root curve?
Curved grade = 10 × √(raw score). A 72 becomes 84.9, a 49 becomes 70 — it lifts struggling scores the most while barely moving high ones (a 90 becomes 94.9). That gentleness at the top is why it's the most popular curve in US classrooms.
How does a flat curve work?
The same number of points is added to every student — typically enough to bring the highest score to 100. Simple and transparent, but it helps top students exactly as much as struggling ones, so it doesn't compress the grade gap.
What is scaling to the highest score?
Every score is divided by the best score in class: if the top student got 88%, everyone's score is multiplied by 100/88. The logic: if nobody reached 100, the test was harder than intended.
Is curving grades fair?
It depends on why the scores were low: a genuinely too-hard test is a fair reason to curve; a class that skipped studying is not. Most teachers curve when the test, not the effort, was the problem — and always apply one method to the whole class.