Strong relationships between one-time serum testosterone (T) measures and sexual activity have been demonstrated in cross-sectional analyses of adolescent males. However, a subsequent longitudinal study of a separate adolescent sample, using semiannual plasma measures, failed to replicate cross-sectional findings. The present study reexamined the relationship between T and sexual activity using more frequently collected measures of salivary T and behavior. Saliva samples and weekly behavior checklists were collected over a 2-year period from the same panel sample of males on which semiannual analyses were based. Saliva samples representing measures approximately every fourth week of study participation were selected for T assay. The association between these monthly salivary T measures and weekly reports of incidents of sexual activity was assessed with repeated measures analyses. Higher levels of salivary T were significantly associated with an increasing hazard of coital initiation, and with more frequent coital and noncoital activity. Within-individual change analyses demonstrated that increases in salivary T were associated with increased sexual activity. Comparisons of models that varied specimen type, schedule of T measurement, and type of behavioral report indicated that all three factors contribute to the varying magnitude of hormone/behavior relationships that are demonstrated. These findings are consistent with a biosocial model of adolescent sexual development that pubertal changes in T are a causal factor in the timing of sexual initiation and the frequency of activity during adolescence.