Do babies really look more like their fathers?

It's the most reliable line in any delivery room: he has your nose. Usually aimed at the father, usually within the first hour. The claim that newborns actually do favor their dads has a real scientific pedigree — one famous study, published in Nature — and a less famous afterlife in which the finding mostly fell apart. Both halves deserve telling.

1995: the study everyone still quotes

Nicholas Christenfeld and Emily Hill at UC San Diego published “Whose baby are you?” in Nature in December 1995. The design was elegant: show volunteers photos of children alongside candidate parents and ask them to match. Per New Scientist's contemporaneous report, 122 volunteers judged photo sets of 24 children — and matched 1-year-olds to their fathers 50% of the time, well above the one-in-three chance baseline. Mothers got no such advantage, and the father effect vanished by age ten. The evolutionary story wrote itself: babies advertise paternity to keep fathers invested.

1999: the replication that didn't

Four years later, Serge Brédart and Robert French at the University of Liège re-ran the design with children aged 1, 3, and 5 and published the result in Evolution and Human Behavior00047-6). Their abstract is unusually blunt: contrary to Christenfeld and Hill, children “do not appear to resemble their fathers significantly more than their mothers,” and any parent-child resemblance, while better than chance, is “far from overwhelming.” Same method, opposite answer. In small-sample psychology, that's less a scandal than a Tuesday — but it means the 1995 number should never be quoted alone.

2000: the twist about who says it

The follow-up literature found something sturdier than the resemblance itself: the claiming of it. Work covered by Times Higher Education in 2000 reported that mothers and their families insist newborns look like the father far more often than neutral judges agree — a bias with an obvious evolutionary reading, since paternity can't be seen but paternal investment can be encouraged. The delivery-room chorus, in other words, is real. It's just not evidence about the baby's face.

Where that leaves the claim

One positive study, one failed replication, and a robust bias in who does the claiming: the honest verdict is unsettled, leaning folk belief. Real resemblance is dealt at conception from thousands of variants across both genomes — which is also why no app can predict a face, only imagine a plausible one. If you want to see one such possibility rendered while the person is still across the table from you, that's the product we make. We rank the alternatives honestly too.