Free Genetics tool
Free Genetics toolEye Color Calculator
The Eye Color Calculator calculates statistical probabilities of child eye color (brown, green, blue) based on parental and grandparental phenotypes. Most of the visible variation in human eye color traces back to a single switch (rs12913832 near OCA2) where one allele keeps pigment production up and the other turns it down. Give both parents' genotypes and this calculator crosses them the way any single-gene trait is crossed, showing the chance of a brown- or blue-eyed child and the genotypes that produce each.
Eye Color Calculator: result
| A | G | |
|---|---|---|
| A | AA → brown | AG → brown |
| G | AG → brown | GG → blue |
This is the single strongest variant, not the whole story: other loci (OCA2 itself, TYR, SLC24A4, IRF4 and more) shift shades toward green and hazel and adjust expression, and forensic tools such as IrisPlex use six markers together rather than one.
At a glance
- What it does: Crosses both parents' rs12913832 genotypes to give the chance of a brown- or blue-eyed child.
- Who it is for: For parents with direct-to-consumer DNA data and students working the classic recessive example.
- Typical result: Returns brown and blue percentages, the genotype mix behind them, and whether a blue-eyed child is possible.
How to use the eye color calculator
- Find each parent's rs12913832 genotype from direct-to-consumer DNA data if available: the options are AA, AG and GG. prefill →
- Select Parent 1's genotype first; AA and AG both read as brown-eyed, GG as blue-eyed under this model. prefill →
- Select Parent 2's genotype the same way. The calculator treats the SNP exactly like any single-gene cross.
- Read the two chance rows: brown-eyed probability, blue-eyed probability, and the genotype mix that produces them.
- Check the 'blue-eyed child possible?' row. It is the question two brown-eyed parents actually ask.
Eye Color: the formula & its source
One locus, two alleles, ordinary Mendelian arithmetic. Each parent contributes one allele, so the cross is a 2 × 2 grid, and the phenotype map is dominant-recessive: any A allele is brown-associated, only GG is blue-associated.
P(blue) = P(G from parent 1) × P(G from parent 2) → AG × AG gives 1/4 GG
The calculator enumerates the four cells directly, which keeps heterozygote-heavy crosses (AG × AG = 25% blue) and carrier-free crosses (AA × AA = 0% blue) honest without special cases.
| Symbol | Meaning | Unit | Typical range |
|---|---|---|---|
dad | Parent 1 rs12913832 genotype | genotype | AA, AG or GG |
mom | Parent 2 rs12913832 genotype | genotype | AA, AG or GG |
A | allele associated with higher iris pigment | allele | one copy per parent |
G | allele associated with reduced pigment | allele | one copy per parent |
Eye Color: worked example
Worked example: AG × AG (the defaults). Gametes are A and G from each parent: AA (25%), AG (50%), GG (25%). Under the dominant model, AA and AG are brown-associated, so the odds are 75% brown-eyed, 25% blue-eyed per child. This is the classic answer to the question in the guide: two brown-eyed carriers, one blue-eyed chance in four. Contrast AA × AA (100% brown, no G allele anywhere) with GG × GG (100% blue).
Example B: blue-eyed parents
Example B: blue-eyed parents. Choose GG for both. The grid produces only GG, so the brown-eyed chance reads 0% and the blue-eyed chance 100% under this single-SNP model. Families do report brown-eyed children from two blue-eyed parents, and those cases involve variants outside rs12913832. The page stays honest by reporting what the model allows rather than what any particular pedigree might contain.
The single switch this page crosses
Most of the visible brown-versus-blue variation in human eye colour traces to one position: rs12913832 in a regulatory element near OCA2. The calculator treats it like any other two-allele locus. Each parent contributes one allele, the grid fills four cells, and the phenotype map is dominant-recessive: any A allele is associated with brown eyes, while only GG reads as blue.
That map makes the arithmetic identical to a textbook single-gene cross. Two heterozygous parents give 25% GG, the classic answer to the question two brown-eyed parents actually ask. Genotype probabilities also frame what prediction can deliver: with both parents genotyped the numbers are exact within the model, whereas predicting from an unknown sample would need population allele frequencies.
The output answers the practical questions in order. A brown-eyed chance row and a blue-eyed chance row come first, each backed by the genotype mix that produces it, so a 75/25 split shows its three cells rather than hiding them. Below that, the blue-eyed-child-possible row states the question two brown-eyed parents actually ask, in one word. AA and AG both read as brown-associated under this model, which is why two brown-eyed carriers can still produce a blue-eyed child one time in four, and why the possible row flips to no only when at least one parent carries no G allele at all.
From genotype to iris pigment
The A allele sits in a HERC2 enhancer that keeps OCA2 expression in the iris high, so melanin production stays up and the stroma fills with pigment. The G allele reduces that enhancer's activity, less pigment reaches the stromal cells, and the shorter, sparser melanin bodies scatter light that reads as blue. The same region was pinned down by the association studies cited at the bottom of this page.
Phenotype and genotype do not always line up. Some AG individuals have blue eyes, some haplotypes read differently across ancestry groups, and eye colour itself can shift during the first years of life as melanin accumulates in infants. The calculator reports the model's answer, which is a probability about a variant rather than a photograph of a child.
Genotyping both parents is what makes the arithmetic exact, and it is also the boundary of what the page claims. With two genotypes in hand there is nothing left to estimate: the four cells are the answer. Predicting from a crime-scene sample or an unknown parent is a different problem, because it needs population allele frequencies to fill the missing side, which is why forensic eye-colour tools report likelihood ratios rather than certainties. The distinction matters when a result gets quoted back to someone as fact.
Assumptions
One locus, two alleles, ordinary segregation, and a dominant-recessive phenotype map: that is the whole model. Both parents are assumed to be genotyped correctly at rs12913832, each passes one allele with probability one half, and the four grid cells are equally likely. Alleles are treated as independent draws, so nothing models linkage to nearby variants carried on the same haplotype.
The brown and blue labels are treated as exhaustive outcomes of this variant, and no account is taken of the other loci that shape shade. Population-specific effects, age-related change and the possibility of modifying variants are all outside the calculation.
The model also freezes a trait that moves. Eye colour can shift through the first years of life as melanin builds up in the iris, so a genotype called in infancy and a phenotype scored at age five may not agree even without any modelling error. Ancestry shifts the picture too: the same genotype does not carry identical effects across populations, and green, hazel and intermediate shades sit outside the two labels this page prints.
Biological limits
Real eye colour is polygenic. Beyond rs12913832, variants in TYR, SLC24A4, IRF4 and others adjust shade and account for green, hazel and intermediate colours that a brown-versus-blue split does not represent at all. Prediction models used in research therefore combine several markers rather than one, and even they report likelihoods rather than certainties.
Two GG parents produce only GG under this model, yet reported families do contain brown-eyed children from blue-eyed parents, through variants the single SNP does not capture. The page also makes no medical, diagnostic or forensic claim: results are for understanding inheritance patterns, and any real-world determination comes from a phenotyped exam or a validated multi-marker panel.
What your eye color result means
The percentages are per child, independent each time. A couple with 25% odds can have three brown-eyed babies before a blue-eyed one appears, with no contradiction.
Genotype probabilities like these also frame what eye-color prediction can and cannot do: with both parents genotyped the arithmetic is exact, but predicting from a crime-scene sample or an unknown parent needs population allele frequencies, which is why forensic eye-color tools report likelihood ratios rather than certainties.
How to read the output:| Result range | What it means | What to do next |
|---|---|---|
| 75% brown, 25% blue | both parents are AG carriers | The classic recessive case, one chance in four |
| 100% brown | at least one AA parent with no G pairing | No blue-eyed child in this model |
| 100% blue | both parents are GG | Only G alleles are available to pass |
| Blue possible: yes | both parents carry a G | Check the genotype mix row for the exact share |
Common mistakes with the eye color
- Reading the percentages as a prediction of one specific child: Each conception is an independent draw, so 25% stays 25% no matter how the first children came out.
- Expecting this model to place green or hazel eyes: Those shades come from other loci outside this model; the calculator only sorts brown-associated against blue-associated genotypes in its two rows.
- Treating a DNA-data genotype as a certainty about eye colour: Genotype and phenotype can disagree at this SNP, so the output is a probability from the model, not a diagnosis.
- Using the result for any clinical or forensic purpose: Prediction in those settings needs multi-marker panels and population frequencies, none of which this single cross contains.
Limits of the eye color calculator
This is a deliberately simplified single-SNP model. Real eye color is polygenic: a handful of common variants plus rare ones, with effects that vary across populations; green and hazel shades are not represented by the brown/blue split at all.
Phenotype and genotype can disagree: some AG individuals have blue eyes, some GG-adjacent haplotypes read differently across ancestry groups, and eye color itself can shift during the first years of life as melanin accumulates. Nothing here is suitable for medical, diagnostic or forensic use.
Research note from Dr. Dinesh A. Nagegowda, PhD: While introductory biology often presents human eye color as a simplistic single-gene dominant/recessive model (brown dominant over blue), modern molecular genetics confirms it is a polygenic trait governed by at least 16 distinct loci. The primary driver is the OCA2-HERC2 locus on chromosome 15, where a single nucleotide polymorphism in the HERC2 intron acts as an enhancer regulating OCA2 expression and melanosome melanin packaging.
Sources
- Sturm et al. (2008), HERC2 rs12913832 association with human eye colour, AJHG 82:424-431 (used for: the SNP-to-eye-colour association this calculator crosses)
- Eiberg et al. (2008), blue eye colour and the OCA2 region, Human Genetics 123:177-187 (used for: OCA2-region evidence behind the brown and blue categories)
- IrisPlex multi-marker prediction system (Walsh et al., 2011) (used for: why multi-marker models are needed beyond one SNP)
- Liu et al. (2010), digital quantification of human eye colour, PLoS Genetics (used for: measurement of eye colour as a graded trait rather than two bins)
Cite This Calculator
For laboratory course reports, academic papers, or institutional protocols (APA format):
Eye Color: frequently asked questions
Can two brown-eyed parents have a blue-eyed baby?
Yes. If both are AG (carriers), each passes G with 50% probability, so 25% of children are GG. The calculator's default cross shows exactly that case, which is why eye color was one of the first traits recognised as recessive in humans.
Is eye color controlled by one gene?
One variant does most of the heavy lifting, but not all of it. rs12913832 near OCA2 explains a large share of brown-vs-blue variation, while TYR, SLC24A4, IRF4 and others adjust shade and account for green, hazel and intermediate colors, which is why prediction models use multiple markers.
What does the A allele actually do?
It sits in a HERC2 regulatory element that drives OCA2 expression in the iris, keeping melanin production high. The G allele reduces that enhancer's activity, less pigment reaches the stromal cells, and shorter, sparser melanin bodies scatter light that looks blue.
Do blue-eyed parents ever have a brown-eyed child?
Under this single-SNP model, no. Two GG parents can only make GG. Rare brown-eyed children from blue-eyed parents do occur in reported families, through variants outside rs12913832 and effects this simple cross does not model.
Found a number that does not match its source? Report a mistake. Corrections are published in place and the page date is updated.
Found a number on this page that doesn’t match its source? Email the correction →.
Dr. Dinesh A. Nagegowda, PhD: Plant & Molecular Biology Editor
Dinesh is a plant molecular biologist and Chief Scientist at CSIR-CIMAP whose research runs through functional genomics, transcriptomics, metabolomics, and biosynthetic pathways in plants. He writes and…