Hazel eyes are neither simply dominant nor simply recessive. The trait is polygenic, meaning at least 16 genes shape iris color, and hazel sits on a continuous spectrum rather than in a two-box dominant-recessive system. Two blue-eyed parents almost never produce a hazel-eyed child, which shows some constraint exists. But two brown-eyed parents can produce hazel-eyed children roughly 1 time in 16. The old classroom model that assigns one gene pair to eye color is a useful fiction, and it mispredicts hazel more than any other eye color.
The mechanism: why hazel sits between brown and blue
Iris color is not pigment painted onto the eye. It is a two-layer optical system, and hazel is what happens when the layers disagree.
The back layer, the iris pigment epithelium, is always densely packed with melanin in every human eye. The front layer, the iris stroma, is where variation lives. Its color depends on how much melanin the melanocytes there produce and how densely they pack it. Blue eyes have almost no stromal melanin. Brown eyes have a lot. Hazel eyes sit in the middle band, with moderate stromal melanin distributed unevenly.
That uneven distribution is the key mechanism. In hazel eyes, melanin concentrates around the pupil and thins toward the outer iris. Light passing through the thinned outer stroma scatters off the back layer and returns as blue-green. The same Rayleigh scattering that makes the sky blue makes the iris rim look blue. The pigmented center reads gold or brown. One eye, two colors, one pigment.
In plain terms: hazel is not a color choice. It is a light-scattering effect layered over a partly pigmented filter.
The gene with the largest single effect is HERC2, which sits on chromosome 15 and controls nearby OCA2. A common variant in HERC2 acts like a dimmer switch on OCA2 expression. When the switch is turned down, less melanin is built in the stroma and eyes trend lighter. When it is turned up, eyes trend darker. This single switch explains roughly 75 percent of the blue-versus-brown variation in European-ancestry populations. It does not explain hazel on its own, because hazel needs that middle setting plus variation from other genes.
What the genes actually do: a polygenic model
More than 16 genes have now been tied to eye color, and the list keeps growing. HERC2 and OCA2 carry the largest weight. Others with confirmed effects include TYR, TYRP1, SLC24A4, SLC45A2, IRF4, and ASIP. Each contributes a small push toward more or less melanin.
This is why hazel behaves like a threshold trait, not a switch. Imagine a dial from 0 to 100. Below roughly 20 you get blue. Above roughly 70 you get brown. The 20-to-70 band produces green and hazel, and where you land inside that band depends on the combined output of many genes.
Because the band is wide, hazel is genetically "unstable" in families. Two hazel parents can land children anywhere from blue to brown, and hazel itself appears in maybe a third to half of their children depending on the underlying genotypes. Two brown parents with hidden light-eye variants can still produce a hazel child. The classic Punnett square cannot capture this because it assumes one gene with two alleles.
A useful correction: the old model treats brown as dominant over blue and green as a simple blend. In reality, no eye color allele is dominant in the Mendelian sense. "Dominant" survives as a convenient label for the tendency of darker variants to show up, not as a genetic law.
Quantified context: how common is hazel, and where
Hazel is rare globally but common in specific populations. Worldwide, roughly 5 percent of people have hazel eyes, though estimates range from 3 to 8 percent depending on how the boundary with green and light brown is drawn.
| Population | Estimated hazel share | Notes |
|---|---|---|
| Worldwide | About 5% | Range 3-8% depending on classification |
| United States | About 5-8% | Second-rarest common color after green |
| Europe overall | About 5-10% | Higher in central and western regions |
| United Kingdom | About 5% | Often grouped with green in surveys |
| North Africa and Middle East | 10-15% | Frequently overlaps with light brown |
| Sub-Saharan Africa | Under 1% | Dark brown dominates |
| East Asia | Under 1% | Brown dominates almost completely |
The classification problem is real. Many surveys lump hazel with green or with light brown, which is why published figures disagree. A study that treats hazel as a separate category will report lower numbers than one that merges it.
For comparison, brown eyes account for roughly 79 percent of people worldwide, blue about 8 to 10 percent, and green about 2 percent. Hazel and green together make up a small minority, and their boundary is the blurriest in the field.
Variations and edge cases that break the simple story
Hazel is the eye color most likely to be misclassified, and several mechanisms explain why.
Lighting and pupil size change the read. A hazel iris photographed in bright sun can look green, and the same eye in dim light can look brown. The pupil widens in low light and covers the pigmented center, leaving more of the blue-green rim visible. This is not illusion; the optical mix genuinely shifts.
Hazel can change across a lifetime, especially in childhood. Many children born with blue eyes darken to hazel or brown by age 3 to 5 as stromal melanin production ramps up. A smaller shift can occur through adolescence. Adult eye color is largely stable after the mid-20s, but slow darkening with age is documented.
Heterochromia is a separate phenomenon. True heterochromia iridum, where the two eyes differ, occurs in under 1 percent of people and is usually harmless. Sectoral heterochromia, where one iris has a wedge of different color, produces a "hazel-like" patchwork that is not the same as true hazel.
Central heterochromia is the edge case most often confused with hazel. In central heterochromia, a distinct ring of gold or brown surrounds the pupil and the rest of the iris is a clearly different color with a sharp boundary. Hazel blends gradually; central heterochromia has a hard edge. Many people who call their eyes hazel actually have central heterochromia over a blue or green base.
Rare syndromes matter for safety. Waardenburg syndrome can produce striking blue or heterochromic eyes alongside hearing loss, and it is worth flagging because eye color change paired with hearing problems is a medical signal, not a cosmetic one. Pigment dispersion syndrome, more common in lighter irises, can cause pigment to shed into the eye and raise glaucoma risk.
Red-flag warnings that warrant an eye exam:
- A change in iris color in adulthood, especially in one eye only
- A new dark spot or growing pigmented lesion on the iris
- Eye color change accompanied by pain, redness, or vision loss
- Heterochromia that appears suddenly rather than being present from birth
- Eye color change together with hearing difficulty or a family history of Waardenburg syndrome
Separating established science from hypothesis
The polygenic model is established. Genome-wide association studies, including large meta-analyses published from 2009 onward, consistently identify HERC2, OCA2, and a growing list of secondary genes. The heritability of eye color is high, estimated above 0.9 in twin studies, meaning genes explain most of the variation.
What remains hypothesis is the precise mapping of genotype to hazel. Researchers can predict blue versus brown with reasonable accuracy from HERC2 and OCA2 alone. Predicting hazel versus green versus light brown is far less reliable because the contributing variants are many and small, and because environmental and developmental factors modulate the final shade.
Another open question is whether hazel is a distinct biological category or a descriptive label on a continuum. Most geneticists treat it as descriptive. That matters for anyone using a consumer DNA test to predict a child's eye color: the predictions are probabilistic and often wrong at the hazel-green boundary.
The number that reconciles the competing claims: HERC2 explains about 75 percent of blue-brown variation, the full polygenic score explains more, but no current model assigns a specific probability to "hazel" with clinical confidence. Treat any single-gene hazel prediction as a rough guess.
What this means for predicting a child's eye color
Parents often want a straight answer, and the honest one is probabilistic.
| Parent eye colors | Chance of hazel child | Why |
|---|---|---|
| Hazel x hazel | Roughly 30-50% | Wide genetic band, many combinations land in hazel |
| Hazel x brown | Roughly 15-25% | Brown parent may carry light variants |
| Hazel x blue | Roughly 10-20% | Blue parent contributes low-melanin variants |
| Brown x brown | Roughly 6% | Both must carry hidden light variants |
| Brown x blue | Roughly 5-10% | Depends heavily on the brown parent's second allele |
| Blue x blue | Under 1% | Almost no melanin-building variants to combine |
The 1-in-16 figure for two brown parents comes from the classic recessive model and is a rough floor, not a precise probability. Real families show more hazel than that simple model predicts, because the polygenic score shifts the odds.
For parents who want concrete action rather than prediction:
- Do not buy a DNA test expecting a reliable hazel prediction; the models are calibrated for blue versus brown.
- Photograph a child's eyes in consistent daylight to track real change rather than guessing.
- Expect eye color to settle by about age 5, with minor shifts into the teen years.
- See an eye doctor if color changes after adulthood or appears in only one eye.
- Remember that hazel carries no health disadvantage on its own; lighter irises do carry modestly higher UV sensitivity.
Limitations and how to read eye color claims
Two limits deserve honesty. First, eye color classification is subjective, so percentages across studies are not strictly comparable. Second, genetic prediction of hazel is still an active research problem, and consumer tests overstate their confidence.
The most defensible statement is this: hazel is neither dominant nor recessive in the classical sense. It is a polygenic threshold trait shaped mainly by HERC2 and OCA2, modulated by at least a dozen other genes, and expressed through light scattering over partial pigment. That mechanism explains why hazel runs in families without following simple ratios, why it can look green in one light and brown in another, and why the old dominant-recessive framing fails specifically at the hazel boundary.