Given the complexity of their work, street-level bureaucrats rely on their professional networks to access implementation resources and information. Despite the acknowledged importance of these networks, little research exists on how network structure and composition influence frontline performance. This study analyzes a unique data set that includes the professional networks of more than 420 teachers in 21 public schools along with 3 years of administrative data on student test scores and student demographics. Using value-added models derived from the student test data, objective measures of teacher performance were calculated. The results suggest that street-level performance is influenced by both network structure and composition. Thus, the actions of street-level workers are not independent responses to individual dilemmas, but rather are developed and shaped by specific features of the social structure in which the individual bureaucrat is embedded.