Similarity in alcohol consumption among adolescent friends could be caused by the influence of friends or by the selection of friends who consume similar levels of alcohol. This article aims to disentangle influence and selection processes while specifically examining changes over time in these processes and possible differences between reciprocal and nonreciprocal friendships. The design was longitudinal with four observations (Time 1-Time 4 [T1-T4]). Data consisted of a longitudinal sample of 1,204 Finnish adolescents in 10 junior high schools. The main measurements were adolescents' friendship networks and alcohol consumption. For three successive periods, T1-T2, T2-T3, and T3-T4, actor-based models for the co-evolution of networks and behavior were analyzed (Mage: T1 = 13.6 years, T2 = 14.6 years, T3 = 15.6 years, T4 = 16.1 years). Selection, as well as influence processes, played an important role in adolescent alcohol consumption. Influence was found during the first period (T1-T2), whereas support for selection was found during the last two periods (T2-T3 and T3-T4). The strength of influence and selection processes did not differ for reciprocal and nonreciprocal friendships. The impact of selection and influence processes changed over time such that influence was only present during early adolescence, whereas selection was present during mid-adolescence. During early adolescence, youngsters would benefit from learning to resist social influence. Alcohol-consumption prevention programs targeting mid-adolescence should consider peer selection processes. These findings stress the importance of considering changes over time in future practice and research.