Strong cross-sectional relationships between testosterone (T) and sexual activity have been found for human adolescent males, even when level of pubertal development is controlled; however, it has not been demonstrated that sexual behavior increases over time as a consequence of hormonal changes. The purpose of this paper was to extend previous cross-sectional findings by demonstrating that sexual behavior is initiated and increases within individuals over time as a direct function of changes in T levels, rather than indirectly through pubertal development. Analyses are based on a 3-year panel study of about 100 7th- and 8th-grade adolescent boys. Pubertal development and changes in pubertal development were significantly and positively related to changes in sexual ideation, noncoital behavior, and to the transition to sexual intercourse. Testosterone levels at study entry were significantly related to coital status at Time 1 and also predicted the transition to nonvirgin status. However, changes in hormone levels during the 3 years of study participation did not predict changes in ideation or noncoital sexual activity, regardless of the period of change or the lag time to effects. The panel data suggest, in contrast to earlier cross-sectional results, that change in pubertal development is related to the initiation of sexual activity because of its social stimulus value, and not solely because it is determined by, or is a proxy for, changing hormone levels. Possible causes of the different findings in the cross-sectional and panel data are examined, and the implications and meaning of hormone/behavior relationships based on single measures are discussed.