We study how different network targeting mechanisms affect the direct and spillover impacts of a school-based health intervention in Zambia. Across 133 schools, we randomly assigned whether the intervention was administered to i) a random selection of students; ii) most central students; or iii) most central students and an additional friend. Our results indicate that the two-hour intervention had broad positive impacts on student sexual and reproductive health behaviour and knowledge. The intervention increased HIV testing rates, increased the likelihood that students visited a health facility for family planning advice over the past 6 months, and increased whether students were aware of any family planning methods. We also find evidence that the intervention had positive spillover effects on the knowledge and behaviour of students that did not receive the program. Our findings also suggest that spillovers were largest when the intervention was targeted to central students (whether or not sampled with an additional friend).