Imagine a test for a rare condition that affects 1 in 100 people. The test is 99% accurate — it correctly identifies 99% of people who actually have the condition, and correctly clears 99% of people who don't. You take it. It comes back positive. How worried should you be?
Most people's instinct is “99% worried.” The real answer, worked out properly, is about 50%. That gap is exactly what Bayes' theorem exists to close — and once you see why, you'll never read a positive test result the same way again.
The two questions that sound the same but aren't
“How accurate is this test?” is really two separate questions wearing one word. There's sensitivity — of everyone who has the condition, what fraction does the test correctly flag? And there's specificity — of everyone who doesn't have it, what fraction does the test correctly clear? A test can score 99% on both and still leave you with a genuinely uncertain result, because neither number tells you what you actually want to know: given a positive result, what's the chance you have the condition?
Working it out with 10,000 people
The cleanest way to see this isn't algebra — it's counting. Imagine testing 10,000 people, where 1% (100 people) actually have the condition.
- Of the 100 who have it, the test correctly flags 99 (sensitivity 99%) and misses 1.
- Of the 9,900 who don't have it, the test incorrectly flags 1% of them — 99 people — as positive anyway (specificity 99%, so a 1% false-positive rate).
Add up everyone who tested positive: 99 true positives plus 99 false positives, 198 people in total. Of those 198, only 99 actually have the condition — exactly half.
10,000 people tested: the 0.99% who are truly positive and the 0.99% who are falsely positive come out almost identical, because the rare condition (1%) means the much larger healthy group still produces just as many false alarms.
Try it yourself — drag the slider to change how rare the condition is, and watch how a “99% accurate” test's real reliability changes with it.
1.0% of people have the condition — 100 out of 10,000 tested.
99
true positives
99
false positives
50%
chance a positive is real
A test that's 99% accurate on paper is only 50% reliable in practice at this prevalence — because 99 healthy people are being flagged positive for every 99 genuinely sick ones caught.
The formula behind the counting
Bayes' theorem is the formal version of exactly that counting argument — it lets you update a starting probability once new evidence arrives:
P(condition | positive) = [P(positive | condition) × P(condition)] ÷ P(positive)
The numerator is the 99 true positives (99% sensitivity × 1% prevalence). The denominator, P(positive), is everyone who tests positive for any reason — true positives and false positives combined, which is where the 198 comes from. The rarer the condition, the more the denominator gets padded out with false positives, and the harder a single positive result has to work to actually mean something.
Why this genuinely matters, not just as an exam topic
This isn't a trick question designed to catch students out — it's the actual reasoning clinicians use every day, formalised as “positive predictive value.” It's why a doctor doesn't diagnose off one positive screening test alone when the condition is rare, and why population-wide screening programmes are debated so carefully — a highly accurate test, applied to millions of healthy people, can still generate an enormous number of false alarms simply because the healthy population is so much larger than the affected one. If probability and statistics are something you or your child need explained with real reasoning behind the formula rather than rote memorisation, that's exactly what our GCSE maths tutoring is for, and you can see the full learning pathway here.
Frequently asked questions
Doesn't a 99%-accurate test mean I'm 99% likely to have the condition if it's positive?
No — and this is the single most common misreading of any diagnostic test. 99% accurate describes how the test behaves given your true condition (it correctly flags 99% of people who have it, and correctly clears 99% of people who don't). It says nothing directly about the reverse question — how likely you are to actually have the condition given a positive result — and those two numbers can be wildly different when the condition itself is rare.
So is the test useless?
Not at all — it's doing real work. Before testing, your chance of having the condition was 1%. After one positive result, it jumped to 50%. That's a genuine, large update; it's just not the same as certainty. In practice, doctors respond to exactly this gap by ordering a second, different test — two independent positives together push the probability up into the high 90s.
Where else does this exact reasoning apply?
Anywhere a test screens for something rare: airport security screening for rare threats, spam filters flagging rare-but-real emails, fraud-detection systems flagging rare-but-real fraudulent transactions, even a smoke alarm in a house that rarely actually catches fire. In every case, a low base rate means false positives can outnumber true positives even when the detector itself is highly accurate.
Reader ratings & reviews
Rate this article across a few different things — your rating helps other readers, and helps us improve.
Was this article helpful?
Tell us what you think — a correction, a question it left unanswered, or a topic you'd like covered next.
About the author
Sudershan Soni
Founder & Lead Tutor at Mostak Services — an MSc-qualified Mathematics, Science, Computer Science & STEM tutor with 20+ years of professional experience, teaching students from 11+ and GCSE to A-Level and beyond, online worldwide.
Read full profile