Most of us would acknowledge that family members often resemble one another, particularly in the face. Indeed, humans are good at picking out pairs of close relatives amongst the faces of unfamiliar adults. We are also more likely to trust and help hypothetical partners whose faces have been subtly manipulated to resemble our own in computer experiments. Furthermore, in natural populations perceived facial similarity can have important consequences in terms of whether fathers invest in or discriminate against their putative children. At what point in the lifespan does the relevant facial information become discernible? This is important to resolving whether there are evolutionary pressures for paternity to be concealed (to avoid withdrawal of care or even infanticide, if a male perceives he is unrelated to the child) or, alternatively, for advertising one’s paternity in order to encourage parental investment. Results of studies examining the faces of newborns have been inconclusive, but there is evidence that by mid-childhood individuals can, on average, be matched to their parents.