The authors of this study tested a selection-influence-de-selection model of depression. This model explains friendship influence processes (i.e., friends' depressive symptoms increase adolescents' depressive symptoms) while controlling for two processes: friendship selection (i.e., selection of friends with similar levels of depressive symptoms) and friendship de-selection (i.e., de-selection of friends with dissimilar levels of depressive symptoms). Further, this study is unique in that these processes were studied both inside and outside the school context. The authors used a social network approach to examine 5 annual measurements of data in a large (N =847) community-based network of adolescents and their friends (M = 14.3 years old at first measurement). Results supported the proposed model: adolescents tend to select friends with similar levels of depression, and friends may increase each other's depressive symptoms as relationships endure. These two processes were most salient outside the school context. At the same time, friendships seemed to be ended more frequently if adolescents' level of depressive symptoms was dissimilar to that of their friends.