Test Grades: Converting Wrong Answers Into Percentages and Letters

Eleven questions wrong out of 50 is 78% — a C on the standard 90/80/70/60 scale, and a C+ under the plus/minus convention many colleges use. The arithmetic starts with a subtraction, because the number on your paper is what you missed, not what you got right: 50 − 11 = 39 correct. Then divide correct by total and scale to 100: 39 ÷ 50 = 0.78, and 0.78 × 100 = 78%. That is the entire computation. The test grade calculator runs it from any correct/total pair and reports the letter under both scales — a distinction that turns out to matter, as shown below.

There is a faster mental route once you notice what one question is worth. On a 50-question test, each question carries 100 ÷ 50 = 2 percentage points, so wrong answers convert to points off at a fixed rate: 11 × 2 = 22 points gone, 100 − 22 = 78%. That framing also shows how close the call was. Ten wrong would have been 40 ÷ 50 = 80% — a B. One question separated the letters, while an A sat six questions further away at 45 ÷ 50 = 90%.

Scoring from the red marks

Teachers grade by marking errors, so a returned test usually announces the miss count — a "−11" at the top of the page — and leaves the percentage to you. The conversion is the one shown above, and it always passes through the same formula:

score % = (correct ÷ total) × 100, where correct = total − wrong

The slip to avoid is dividing the wrong-answer count by the total. That gives 11 ÷ 50 = 0.22, or 22% — a real number, but it is the share of the test you missed. Your score is its complement: 100 − 22 = 78. The two figures always sum to 100, which doubles as a quick check on your own arithmetic. If you want an independent second opinion, the percentage calculator in its "what percent is one number of another" mode performs the identical division under different labels: part 39, whole 50, result 78.

The wrong-answer chart for a 50-question test

Because every question on an equal-weight 50-question test costs exactly 2 points, the whole grading range fits in one chart. Each row is 100 minus twice the wrong count; letters follow the two scales the calculator implements.

The clean landings are not a coincidence: 2 divides evenly into the 90/80/70/60 cutoffs, so the standard-scale boundaries fall exactly on 5, 10, 15, and 20 wrong. A 50-question test gives you no rounding ambiguity — you are either at 90% or at 88%, never at 89-point-something. Note also how the two letter columns disagree on six of the eleven rows while describing the same paper.

Every test length sets its own price per mistake

This chart works only for tests of exactly 50 equal-weight questions, because the percentage cost of one wrong answer is 100 ÷ total — and that number swings hard with test length. Here is the same performance, three questions wrong, priced across six common test sizes:

Identical mistakes, spread across three different letter grades. A 10-question pop quiz punishes each slip ten times as hard as a 100-question final, which is why short quizzes feel so unforgiving — and why the calculator surfaces "value of each question" as its own output rather than leaving it implicit. One quiet detail from the last row: under the plus/minus scale, 97% is still a plain A, because the A+ band does not begin until 98.

One percentage, two letters — and neither is wrong

The percentage is arithmetic; the letter is policy. The United States has no national grading scale — the Department of Education's National Center for Education Statistics, which the calculator cites, documents that conventions vary by state, district, and individual instructor. The tool therefore offers two published conventions instead of pretending one is universal. The standard scale awards whole letters at 90/80/70/60. The plus/minus scale mirrors the cutoff table published by Columbia University's School of Professional Studies: a full A begins at 93, A− covers 90–92.99, B+ runs 87–89.99, and the bands continue down to C− at 70–72.99, with a single D band from 60 to 69.99 and F below 60.

The practical consequence: the same 88% is a B on one syllabus and a B+ on another, and a 90.0% is either an A or an A−. Cutoffs are applied exactly as published — an 89.99% is a B on the standard scale because it never reaches 90. Whether a teacher chooses to round an 89.5 up is a classroom policy, not a math question, so check the syllabus before assuming either answer.

When questions carry different point values

The wrong-answer shortcut assumes every question is worth the same. Many exams break that assumption, and then you must count points, not questions. Suppose an exam has 40 multiple-choice questions at 2 points each (80 points) plus two short essays at 10 points each (20 points), for 100 possible. Missing 3 multiple-choice questions costs 3 × 2 = 6 points; dropping 5 points across the essays brings the loss to 11 points. That is 100 − 11 = 89 earned, and 89 ÷ 100 = 89% — a B on the standard scale, a B+ on plus/minus. Note that "lost 11 points on a 100-point exam" and "11 questions wrong out of 50" both contain an eleven and produce completely different grades; the denominator and the per-item value decide everything. To grade a mixed-value test, total the points you earned, total the points possible, and enter those two sums as correct and total.

Measuring the climb between two tests

Once a score is a percentage, improvement has two honest measurements — and they are not interchangeable. Going from 78% on one test to 90% on the next is a rise of 12 percentage points (simple subtraction: 90 − 78 = 12). As a percent change it is larger: (90 − 78) ÷ 78 × 100 = 12 ÷ 78 × 100 = 15.38, roughly a 15.4% improvement, because the change is measured relative to the 78 you started from. The percent-change mode of the same percentage calculator linked above handles the division and the sign for you. Saying "I improved 12%" when you mean 12 percentage points understates a genuinely bigger jump.

Everything in this post can be recomputed by hand from two whole numbers — which is exactly why a returned test is worth checking before the grade posts to a transcript. The test-grade tool lives among the education tools on Quanta, next to the class-level calculators this post deliberately stayed out of: what you need on a final, how a curve moves a raw score, how weighted categories combine. And if your school publishes a scale that matches neither convention here — split D bands, a 7-point scale — send us the published table and we will look at supporting it.

Sources

• Columbia University SPS — grades and percentage-to-letter conventions
• Rowan University Registrar — grades and grading
• NCES — High School Transcript Studies methodology
• Khan Academy — finding percentages