Two Brown-Eyed Parents, and the Model That Says 2.7 % Blue
Published 9/10/2026 · 4 min read · Everyday calculators
Antonio Delgado — Traditions & pastimes writer at OneKitly
Symbolism · Calendars
Checked against 3 sources
The model is the one published in textbooks for decades: two loci, brown dominant over blue at the first and green dominant over blue at the second, giving brown for anyone with a brown allele, green for the rest who carry a green one, and blue for those with neither. From an observed eye colour it works backwards to the genotypes that colour is compatible with, weights them, crosses every combination and adds up — so the output is a probability computed exactly, not a simulated draw, and running it twice gives the same answer. Two brown-eyed parents come out at 92.4 % brown, 4.9 % green and 2.7 % blue, which answers the question people actually arrive with. Brown with blue gives 72.5 % brown and 16.5 % blue. Green with green gives 91.0 % green. And blue with blue gives 100 % blue — the one categorical claim in the table, and the one that observation has repeatedly contradicted, because eye colour is polygenic and two loci are a simplification rather than the mechanism.
The classic two-gene predictor is arithmetic, not a draw — and its one confident answer, blue × blue giving 100 % blue, is exactly where real genetics disagrees with it.
Where the model is known to be wrong
Two loci are not the mechanism. Eye colour is polygenic: the pair around OCA2 and HERC2 does most of the work, which is why the simple model gets so much roughly right, but a dozen or more further variants shift the outcome, and the amount of melanin in the iris is a continuum rather than three boxes. The visible consequence is exactly the cell this table is most confident about — blue-eyed parents can and occasionally do have a green or brown-eyed child, which the two-locus arithmetic declares impossible. Read that 100 % as the model speaking, not as biology.
Why the colour changes in the first year anyway
Melanin accumulates in the iris after birth, so many babies born with grey-blue eyes darken over the first six to twelve months and some keep going into the third year. A prediction made before birth is therefore being compared against a colour that is not settled yet, and a parent checking at three months is checking too early. That is not a flaw in the model — it is the reason the question feels unanswerable for so long, and the reason a table of probabilities is the honest form of the answer rather than a single colour.
| Parents | Brown | Green | Blue |
|---|---|---|---|
| brown × brown | 92.4 % | 4.9 % | 2.7 % |
| brown × hazel | 86.3 % | 10.0 % | 3.7 % |
| hazel × hazel | 75.0 % | 19.9 % | 5.1 % |
| brown × green | 72.5 % | 22.6 % | 4.9 % |
| brown × blue | 72.5 % | 11.0 % | 16.5 % |
| green × green | 0 % | 91.0 % | 9.0 % |
| blue × green | 0 % | 70.0 % | 30.0 % |
| blue × blue | 0 % | 0 % | 100 % |
Frequently asked questions
- Does the tool give a different answer each time?
- No. It enumerates every genotype combination the parents' colours allow, weights each and sums the probabilities — there is no sampling anywhere, so the same inputs always produce the same three numbers, and they add to exactly 100 %. A predictor that gives a different answer on a second click is drawing at random, which is a different and much weaker thing.
- Can I use the grandparents' eye colours?
- Yes, and it sharpens the answer where it matters most. A brown-eyed parent is either BB or Bb, and the model has to weight both; a blue-eyed grandparent proves the parent carries a blue allele and removes BB from consideration entirely. That single fact can double or halve the blue figure for a couple. It changes nothing when the parents are already blue-eyed, since their genotype was never in doubt.
- Is grey the same as blue here?
- At these two loci, yes: grey and blue share the same low-melanin genotype, and the model treats them alike. Hazel is handled as a brown genotype with a raised tendency towards green. Those are modelling choices, not findings — they are what makes a five-colour input work with a three-colour output, and they are worth knowing before reading a percentage as though it came from a laboratory.
Articles you may find interesting
All guides →Related tools
This is the classic two-gene teaching model, not a genetic test. Eye colour is polygenic and the model is known to be wrong at its edges; nothing here is medical advice or a basis for any decision.
Sources
Spotted a mistake in this article?