← Blog
12 min read

How to Curve Grades Fairly: 5 Methods for Teachers (with Examples)

A teacher at a desk reviewing a printed spreadsheet of exam scores, with a bell-curve graph and a calculator beside the papers

An exam comes back rougher than you expected — half the class below passing, nobody near the top — and the question lands: should you curve, and if so, how? Learning how to curve grades well is less about rescuing a bad test day than about choosing a method whose winners and losers you can defend to a student, a parent, or a department head. Every curve helps someone and quietly costs someone else, and the five common methods spread that trade very differently.

This guide runs all five grade curving methods — a flat point bump, scaling to the top score, the square-root curve, the bell curve (norm-referenced grading), and dropping a bad question — on one class's raw scores, so you see exactly who gains and who loses. One caveat: the letter grades (A–F), the 90/80/70/60 cutoffs, and GPA here are common US conventions and illustrative only. Your school's scale — and whether curving is even permitted — is set by your own district, so confirm the specifics in your syllabus or faculty handbook.

How to Curve Grades: Start With What It Means

Curving is any systematic adjustment to raw scores after an assessment, and underneath the mechanics sit two philosophies that university teaching centers stress. Criterion-referenced grading measures each student against a fixed standard: did they demonstrate mastery, regardless of how anyone else did? Norm-referenced grading measures each student against classmates: where did they rank? Only the true bell curve is genuinely norm-referenced; the other four adjust scores but still grade everyone against a standard. For how raw scores become letters and weights in the first place, see how grading works.

Our Sample Class: One Set of Raw Scores

Here is a single exam, scored out of 100, for a class of ten. The raw mean is 60, the top score is 80, and five students sit below a 60 passing line — a genuinely tough result worth adjusting.

Student Raw Flat +10 Scale to top (×1.25) Square-root (10√)
Ava 80 90 100 89.4
Ben 76 86 95 87.2
Carla 71 81 88.8 84.3
Dev 66 76 82.5 81.2
Ella 62 72 77.5 78.7
Finn 58 68 72.5 76.2
Gia 54 64 67.5 73.5
Hugo 49 59 61.3 70.0
Ivy 45 55 56.3 67.1
Jae 39 49 48.8 62.4
Class mean 60.0 70.0 75.0 77.0

Read across any row to watch the three point-transform curves pull in different directions. (See percentage to letter grade for how cutoffs vary.) The bell curve and drop-a-question methods don't fit a single column, so each gets its own example below.

Method 1: The Flat (Linear) Point Bump

How it works: Add the same number of points to everyone. Here we add 10, lifting the top score of 80 to a 90.

On our class: Every student gains exactly 10 points. Passes (≥60) rise from five to seven — Finn and Gia clear the line, and Ava earns the class's first A. Hugo, Ivy, and Jae still fail.

Tradeoffs: The flat bump is the most transparent curve and the easiest to defend: everyone is treated identically and the rank order never changes. Its weakness is that it does nothing targeted — it hands 10 free points to your strongest student and still leaves your weakest short of passing.

Method 2: Scale to the Top Score

How it works: Treat the highest actual score as the new 100 and scale everyone proportionally. With a top of 80, the multiplier is 100 ÷ 80 = 1.25, so every raw score is multiplied by 1.25.

On our class: The mean jumps to 75 — but look at the absolute gains. Ava climbs 20 points (80 → 100) while Jae climbs under 10 (39 → 48.8). Because the curve is multiplicative, higher scorers gain more points. Passes rise to eight and two A's appear; Ivy and Jae still fail.

Tradeoffs: Scaling to the top assumes the test, not the class, was too hard, treating the best performance as full marks. The cost is fairness at the bottom: the students who struggled most gain the least. It also makes your whole scale hostage to one student — had your top scorer earned a 95, the multiplier would shrink and the curve nearly vanish.

Method 3: The Square-Root Curve

How it works: Replace each score with 10 times its square root (a 64 becomes 10 × 8 = 80). Because square roots grow fast at the low end and slowly at the high end, this curve lifts low scores far more than high ones.

On our class: This is the mirror image of scaling to the top. Jae gains a massive 23 points (39 → 62.4) while Ava gains just 9 (80 → 89.4). Every student now passes — zero F's — yet nobody reaches an A, because even the top raw of 80 only climbs to 89.4. The class mean, 77, is the highest of the three point transforms.

Tradeoffs: The square-root curve is the most forgiving to struggling students and the best at clearing a failing tail, which is why teachers reach for it after a brutal exam. The price is compression at the top: your strongest students see their advantage shrink, a raw 80 can no longer earn an A, and if telling your best performers apart matters, this curve works against you.

Method 4: The Bell Curve (Norm-Referenced Grading)

How it works: Instead of transforming scores, you fix the distribution of grades in advance and assign letters by rank. A typical quota for ten students: top 20% earn A, next 30% B, next 30% C, next 10% D, bottom 10% F.

On our class, ranked by raw score:

Tradeoffs: This is the only genuinely norm-referenced method here, and it behaves unlike the others. Ava earns an A with a raw 80 purely because she ranked first, while Jae fails by construction — someone had to fill the bottom slot. Raise the whole class's mastery and nothing changes: if everyone scored a 92, one student would still get an F. Because grades reflect rank rather than what a student knows, the bell curve is the most controversial method, for reasons the next section covers.

Method 5: Drop the Worst Question (Reweighting)

How it works: Sometimes the fix isn't a curve at all but a bad item. Suppose question 12 (5 points) was ambiguously worded and only two students got it. You drop it, regrade the exam out of the remaining 95 points, and rescale to 100.

On our class, three students tell the story. Ella and Jae both missed question 12; Ava answered it correctly:

Tradeoffs: Reweighting is the most surgically fair option when a specific question was genuinely flawed, because it targets the students the bad item hurt. But a naive rescale can quietly penalize students who answered the dropped question correctly, as Ava's dip shows. The common fix is to hold harmless — award everyone credit for the dropped item instead of removing it, so Ella and Jae rise while Ava stays put and no one loses ground. This makes sense only when the item failed, not when the class did.

How to Curve Grades Fairly: Match the Method to the Problem

Same ten students, five methods, very different outcomes:

No method is "the fair one." Fairness depends on what went wrong — a hard test, a flawed question, or a class that hasn't mastered the material yet. That last case is a signal to reteach, not to curve.

The Equity Problem With Bell Curves

Of the five, the norm-referenced bell curve draws the sharpest criticism. The Vanderbilt University Center for Teaching and Carnegie Mellon University's Eberly Center for Teaching Excellence both describe norm-referenced grading as scoring students relative to one another rather than to a standard of mastery — so a grade depends on who else is in the room. Assessment researcher Thomas Guskey has long argued that curving pits students against each other, discourages collaboration, and can mask a teaching problem: if a whole class truly mastered the content, a curve still manufactures failures instead of recording that success.

The criterion-referenced alternative — grading against clear, fixed standards — is where much of the field has moved, part of the broader shift toward standards-based grading covered by outlets like Edutopia. None of this makes curving forbidden; it is one philosophy with tradeoffs. Know which method measures mastery and which measures rank before you choose.

Frequently Asked Questions

How do you curve grades?

Pick a rule and apply it to every raw score consistently. The simplest is a flat point bump — add the same points to everyone. Others scale scores to the top mark, apply a square-root curve to lift low scores, assign letters by rank (a bell curve), or drop a flawed question and reweight. Choose the method whose outcomes you're prepared to justify.

What is a bell curve in grading?

A bell curve, or norm-referenced grading, fixes the distribution of letter grades in advance and assigns them by rank rather than by score. A set percentage earns an A, another a B, and so on, so a fixed number of students land at the top and a fixed number fail — no matter how much anyone actually learned. It measures students against each other, not against a standard.

Is curving grades fair?

It depends on why you're curving. Correcting a miscalibrated test or a genuinely flawed question is widely seen as fair. Grading on a norm-referenced curve is more contested: because it ranks students against each other, someone fails by design even if the whole class mastered the material — which is why many teaching centers and researchers discourage it. There's no universal rule.

What is a square-root curve?

It replaces each score with 10 times its square root, so a 64 becomes 80 and a 49 becomes 70. Because square roots rise quickly at low values and slowly at high ones, it boosts struggling students far more than strong ones — effective at pulling the lowest scores back above passing, but it compresses the top and can put an A out of reach for even the highest raw scores.

Should I curve my exam?

Curve when the test misfired — it was too long, ambiguous, or covered material you hadn't taught — because the scores don't reflect what students know. Don't curve to paper over a class that simply hasn't learned the content yet; that's a signal to reteach and reassess. Check your department's grading policy first, since some prohibit curving outright.

What's the difference between norm-referenced and criterion-referenced grading?

Criterion-referenced grading measures each student against a fixed standard of mastery — the same 90 earns an A no matter who else is in the class. Norm-referenced grading (the bell curve) measures each student against classmates by rank, so grades shift with the group. Four of the five methods here still grade against a standard; only the bell curve is truly norm-referenced.

See the Whole Grade Before You Curve

An exam score is rarely a student's whole grade — it's one weighted slice of the course, so the honest question isn't only what a curve does to the test but what it does to each student's course grade. Pick a method from this guide, apply it to your raw scores, then enter the curved percentage into the Grade Calculator alongside your other category weights to watch the weighted grade and letter move — it flags any total that doesn't add up to 100%. (For how those weights combine, see how grade weighting works.) From there, roll finished grades into a term average with the GPA Calculator, or work backward from a target score with the Final Grade Calculator.

Because these tools run entirely in your browser, you can run a real student's numbers during a meeting without anything touching a server.