A 5 year longitudinal study investigated the interrelations of temperament dimensions (negative emotionality, activity, and sociability), perceived social support (friend, signi®cant other, and family support), and depressive tendencies. Temperament (EASTS; Buss, 1991), perceived social support (PSS-R; Blumenthal, Burg, Barefoot, Williams, Haney and Zimet, 1987) and depressive tendencies (a modi®ed version of the BDI) were measured by self-reports in a randomly selected, community-based sample of 302 adolescents at age 15, and depressive tendencies were measured again ®ve years later at age 20. The authors hypothesized a model comprising direct as well as mediated eects between the adolescent predictor variables and depressive tendencies in young adulthood. The structural modelling analyses supported these hypotheses in part, and also indicated some qualitative sex dierences. For both girls and boys, a signi®cant direct as well as mediating role was indicated for the current level of depressive tendencies: depressive tendencies at age 15 predicted depressive tendencies at age 20, also mediating the eects of negative emotionality at age 15, and additionally mediating the eects of a low level of activity in girls and a low level of family support in boys. Furthermore, even when the prior level of depressive tendencies was controlled, direct eects were indicated in boys for a low level of sociability, and in girls for signi®cant other support, which also mediated the eects of a low level of sociability. These results highlight the importance of the sex-dierentiated interplay of both intrapersonal and psychosocial factors in relation to depressive tendencies both during adolescence and from adolescence to young adulthood.