The objectives of this longitudinal study were to predict the tobacco and alcohol use of Indonesian Muslim adolescents from their religiosity and the substance use of friends and network affiliates. At Year 1, there were 996 participants from eighth grade (n = 507, age = 13.4 years) and 10th grade (n = 489, age = 15.4); 875 were followed into the next grade. Friends and network affiliates were respectively identified by mutual selection and social cognitive mapping. Logistic regression analyses revealed that tobacco use was predicted at both Years 1 and 2 by friends' use, whereas network affiliates' use predicted individual use with the exception of girls at Year 2. Boys' alcohol use was uniquely predicted by network affiliates' but not friends' use at both years, whereas girls' use was not predicted by either friends' or network affiliates' use. Involvement with boys was associated with girls' use of alcohol. Although these two types of peer associations may both be important, they may be associated in different ways with tobacco and alcohol use.