This paper examined the association between friendship and academic networks and how the connections these networks have with academic performance and school misconduct differ when comparing three types of classrooms where students were grouped based on their academic ability (i.e., high-, low-, and mixed-ability). The sample was composed of 528 seventh to ninth graders (Mage = 15; 64.1% girls) from 12 classrooms (four in each category of ability grouping) across two waves in five schools in Chile. The effects of academic performance and school misconduct on receiving academic and friendship nominations were examined, as well as the interplay between academic and friendship relationships. Furthermore, the extent to which similarity in adolescents' academic performance and school misconduct contributed to the formation and maintenance of academic and friendship relationships was examined. Sex, socioeconomic status, and structural network features were also taken into account. Longitudinal social network analyses (RSiena) indicated that (1) in high-ability classrooms students chose high-achieving peers as academic partners; (2) in high-ability classrooms students avoided deviant peers (i.e., those high in school misconduct) as academic partners; and (3) academic relationships led to friendships, and vice versa, in both high- and low-ability classrooms. Whereas the interplay of friendship and academic relationships was similar in high- and low-ability classrooms, the formation and maintenance of academic networks unfolded differently in these two types of classrooms.