User retention could significantly affect the results of web-based health interventions, e.g. weight loss. Past research reported that social ties could have positive influence on retention in weight loss program. However, in this study, it is proposed that social capital could be the reason underlying the effects of social ties on retention and health outcomes in web-based health interventions. It is hypothesized that social capital of participants in web-based health interventions affects their health outcomes directly, and indirectly via increasing user retention in the intervention. This research used the survey method, with which we conducted a 10-week-long physician-supervised weight loss program. In total, 48 participants were recruited, and everyone was required to join in an online health community specially designed for this program. The health outcomes were measured in terms of changes in weight, body fat, waist circumference, and body mass index. The empirical results support the hypotheses and show that social capital of participants can enhance health outcomes directly and indirectly through greater user retention in web-based health intervention.