Six pairs of photographs showing human faces of both sexes were presented to 98 women who had to choose the more pleasing one of each pair. Faces within each pair were identical except for a slight difference in complexion. For women not taking oral contraceptives, skin-color preference differed significantly between two groups of subjects classified according to the current phase of their self-reported menstrual cycle: darker male faces were judged more positively by subjects in the phase when the estrogen/progesterone ratio was expected to be high than by those in the phase when this ratio was expected to be low. Female faces evoked no such cyclic response. Users of oral contraceptives showed no cyclic response to either male or female faces. These results suggest a mental mechanism whose inputs are (a) hormonal state, (b) visual identification of the sex of the face being observed, and (c) visual recognition of complexion, and whose output enters into evaluation of male and female faces. Replication with direct measures of hormonal state is recommended.