To test the effectiveness of a social network intervention (SNI) to improve children’s healthy drinking behaviors. A three-arm cluster randomized control trial design was used. In the SNI, a subset of children were selected and trained as ‘influence agents’ to promote water consumption–as an alternative to sugar-sweetened beverages (SSB)–among their peers. In the active control condition, all children were simultaneously exposed to the benefits of water consumption. The control condition received no intervention. Four hundred and fifty-one children (M age = 10·74, SDage = 0·97; 50·8 % girls). Structural path models showed that children exposed to the SNI consumed 0·20 less SSB per day compared to those in the control condition (β = 0·25, P = 0·035). There was a trend showing that children exposed to the SNI consumed 0·17 less SSB per day than those in the active control condition (β = 0·20, P = 0·061). No differences were found between conditions for water consumption. However, the moderation effects of descriptive norms (β = –0·12, P = 0·028) and injunctive norms (β = 0·11–0·14, both P = 0·050) indicated that norms are more strongly linked to water consumption in the SNI condition compared to the active control and control conditions. These findings suggest that a SNI promoting healthy drinking behaviors may prevent children from consuming more SSB. Moreover, for water consumption, the prevailing social norms in the context play an important role in mitigating the effectiveness of the SNI.