Self-reports about 70 masculine and feminine behaviors were collected from 177 undergraduates and their families. These self-report scores were correlated within family pairs to assess family resemblances in masculine or feminine preferences. The obtained family correlations were compared with those predicted by a simple additive-genetic theory, an X- linkage model, and a same-sex identification theory of gender-identity acquisition. The obtained pattern of correlations (mother x daughter=.26, sister x sister=.28, father x son=.33, and brother x brother=.23, all other family correlations near 0) strongly rejects the X-linkage and simple additive-genetic models of gender-identity acquisition. On the other hand, the obtained family correlations are quite consistent with a same-sex theory of gender identification, although, of course, within-family correlations only suggest causal relations, they do not prove them. Other factors such as social class could affect sex role identification as well.