Skip to content
K Knidox Search…
Grades · Teaching

Grading Scale Maker

Build a points-to-letter conversion table for any test total.

The maximum a student can score.
Points needed for an A
36 / 40

Each point is worth 2.50 percentage points, so one mark moves a borderline student noticeably.

Points to letter grade
GradePointsPercentage
A36–4090% and above
B32–3580% and above
C28–3170% and above
D24–2760% and above
F0–230% and above

Cut-offs are rounded up to the next whole mark, so a student needs to actually reach the threshold rather than round into it. On a short test each point is worth a lot — out of 20, a single mark is five percentage points and can cross two grade boundaries.

A grading scale converts raw marks into letters. On a 40-point test using the standard 90/80/70/60 cut-offs, an A starts at 36 points, a B at 32, a C at 28, and a D at 24 — so 23 or below fails.

Build the table once, mark against it

Converting each student's score individually is slow and error-prone. Working out the point thresholds once, before marking, means the whole class can be graded by looking up a number — and it makes the boundaries visible to you in advance, when you can still decide whether they are fair.

Each cut-off is the smallest whole mark that reaches the percentage threshold. On a 40-point test, 90% is exactly 36, so an A starts at 36. On a 35-point test, 90% is 31.5, and since half marks may not exist the threshold rounds up to 32 — a student must genuinely reach the boundary rather than round into it.

Short tests have coarse grades

The most useful thing this table reveals is how much a single mark is worth. Out of 40 points, one mark is 2.5 percentage points. Out of 20, one mark is 5 points — enough to cross an entire grade boundary, and on a plus/minus scale sometimes two.

That is a design problem, not a marking problem. If a single question can move a student from B to A−, the assessment is not fine-grained enough to justify the distinctions the scale is drawing. Either lengthen the test or use fewer grade bands.

Points needed = ⌈(percentage cut-off ÷ 100) × total points⌉

Rounded up to the next whole mark, so a student must reach the threshold rather than round into it. One mark is worth 100 ÷ total percentage points.

Worked example: a 40-point test

Convert each percentage boundary into a mark:

  1. 1
    Note the total points. The test is out of 40, so each point is worth 2.5 percentage points.
  2. 2
    Take the A boundary. 90% of 40 = 36 exactly, so an A starts at 36 points.
  3. 3
    Work down the bands. 80% of 40 = 32 for a B, 70% = 28 for a C, 60% = 24 for a D.
  4. 4
    Round up where it is not exact. On a 35-point test, 90% is 31.5, which becomes 32 — the student must reach the boundary, not round into it.
  5. 5
    Set the top of each band. A band runs up to one mark below the next threshold, so a B is 32 to 35 and an A is 36 to 40.
  6. 6
    Sanity-check the granularity. With 2.5 points per mark, one question can shift a borderline student a full grade. On a 20-point test it would be worse.

What one mark is worth

The percentage value of a single point, and how many marks separate adjacent letter grades on the standard 10-point bands.

Total pointsOne mark isMarks per grade band
1010.0%1
205.0%2
254.0%2.5
402.5%4
502.0%5
1001.0%10

Rounding is a policy, not a calculation

The scale here rounds thresholds up, which is the stricter and more common convention: 89.9% is not an A. Some institutions round the student's percentage instead, so 89.5% becomes 90% and earns the A. Both are defensible, but they are different rules and they must be stated in advance rather than decided when a specific student lands on the boundary.

Whichever you use, apply it uniformly. The borderline case is exactly where a grading scheme is tested, and deciding it case by case is how a scale stops being a scale.

Two further notes. The 90/80/70/60 cut-offs are an American convention and are far from universal — many systems pass at 50%, and some use entirely different band widths, so adjust to whatever your institution mandates. And plus/minus grades on a short test are usually false precision: distinguishing a B+ from a B on a 20-point paper means claiming that a single mark carries real meaning about a student's understanding, which it rarely does.

Why are thresholds rounded up?
So a student must actually reach the boundary rather than round into it — 89.9% is not an A. Some institutions round the percentage instead, which is equally defensible but a different rule that must be stated in advance.
How many points should a test be out of?
Enough that one mark does not cross a grade boundary. Out of 20, a single mark is five percentage points and can move a student a full grade, which is more precision than one question can justify.
Are the 90/80/70/60 cut-offs standard everywhere?
No. They are an American convention. Many systems pass at 50% and use different band widths, so use whatever your institution mandates rather than assuming these.
Should I use plus and minus grades?
Only on a test long enough to support them. On a 20-point paper, separating a B+ from a B claims that one mark reflects a real difference in understanding, which is usually false precision.
What happens at exact boundaries?
A score at the threshold earns the higher grade — a student on exactly 36 out of 40 gets the A. The band runs from that mark up to one below the next threshold.
Can I use this for weighted assessments?
Convert the weighted result to a percentage first, then read it against the scale. This tool works from a single point total, so combining several weighted components happens before you use it.
Should I adjust the scale after seeing the results?
That is curving, and it is a separate decision. Changing thresholds after marking to reach a desired distribution needs its own justification, and the grade curve calculator handles the arithmetic.