BACKGROUND: Social capital has been considered aetiologically important in schizophrenia but the empirical evidence to support this hypothesis is absent. We tested whether social capital, measured at the neighbourhood level, was associated with the incidence of schizophrenia (ICD-10 F20). MethodWe administered a cross-sectional questionnaire on social capital to 5% of the adult population in 33 neighbourhoods (wards) in South London (n=16 459). The questionnaire contained items relating to two social capital constructs: social cohesion and trust (SC&T) and social disorganization (SocD). Schizophrenia incidence rates, estimated using data from the Aetiology and Ethnicity in Schizophrenia and Other Psychoses (AESOP) study, provided the outcome. We used multi-level Poisson regression to test our hypothesis while controlling for individual- and neighbourhood-level characteristics. RESULTS: We identified 148 cases during 565 576 person-years at-risk. Twenty-six per cent of the variation in incidence rates was attributable to neighbourhood-level characteristics. Response from the social capital survey was 25.7%. The association between SC&T and schizophrenia was U-shaped. Compared with neighbourhoods with medial levels of SC&T, incidence rates were significantly higher in neighbourhoods with low [incidence rates ratio (IRR) 2.0, 95% confidence interval (CI) 1.2-3.3] and high (IRR 2.5, 95% CI 1.3-4.8) levels of SC&T, independent of age, sex, ethnicity, ethnic density, ethnic fragmentation and socio-economic deprivation. ConclusionNeighbourhood variation in SC&T was non-linearly associated with the incidence of schizophrenia within an urban area. Neighbourhoods with low SC&T may fail to mediate social stress whereas high SC&T neighbourhoods may have greater informal social control or may increase the risk of schizophrenia for residents excluded from accessing available social capital.