Social capital is widely cited as benefiting children's school performance, but close inspection of existing research yields inconsistent findings. Focusing on intergenerational closure among parents of children in the same school, this article draws from a field experiment to test the effects of social capital on children's achievement in reading and mathematics. When children were in first grade, their schools were randomly assigned to an after-school family-based intervention that boosts social capital. A total of 52 schools in Phoenix, Arizona, and San Antonio, Texas, containing over 3,000 first graders, participated in the study, with half the schools in each city assigned to the treatment group and half serving as no-treatment controls. Two years later, no differences in third-grade achievement were evident between children who had been in treatment schools versus control schools. By contrast, nonexperimental analyses of survey-based measures of social capital suggest positive effects on achievement, indicating that naïve estimates based on survey measures may be upwardly biased by unobserved conditions that lead to both stronger ties among parents and higher test scores. This article adds to a growing literature that raises doubts about the effects of this type of social capital for achievement outcomes among young children.