A twin/family design was used to explore genetic contributions to personality; to evaluate whether twins and nontwins yield different genetic results; and to test for the presence of contrast effects, the tendency of a rater to contrast one sibling with the other, thereby magnifying existing behavioral differences. The sample consisted of 708 adolescent same-sex sibling pairs from 10 to 18 years of age. Pairs included identical (monozygotic; MZ) and fraternal (dizygotic; DZ) twins, and full siblings in nondivorced families; and full, half, and unrelated siblings in stepfamilies. Mothers and fathers rated the temperament of their children on the EAS Temperament Survey (A. H. Buss & R. Plomin, 1984). Model-fitting analyses revealed significant genetic influences on each of the four EAS dimensions; however, for some dimensions, heritability estimates were significantly greater for twins than for nontwins. Overall, the data were best described by a sibling interaction model, which indicated significant contrast effects.