Background: To identify geographical clusters of suicide in São Paulo, Brazil (2006–2015) and to verify the associations of suicide with sociocultural characteristics of its 96 districts. Methods: Spatial scan test was used to detect the geographical clusters. Correlation and multiple regression techniques were used to estimate the association of socioeconomic and cultural variables with suicide. Results: The mean suicide rate was 4.8/100,000. Three clusters were identified which are as follows: one of increased risk in downtown and two of decreased risk in the South and in the Southeast. The mean suicide rate of the high-risk clustered districts (7.99/100,000) presented significantly higher average incomes per household, higher proportion of formally educated, of no religious affiliation, of recent migrants, of all-times migrants and lower proportion of married. The multiple model selected two independent risk factors – people with no religious affiliation (β = 0.182) and of recent migrants (β = 0.278) – and two protective factors – the proportion of married (β = –0.185) and of total migrants (β = –0.075), which jointly explained 58.4% of the variance. Conclusion: Durkheimian social and cultural risk factors for suicide were confirmed. Compared to a previous study period (1996–2005), suicide rates and geographical clusters remained relatively stable in the subsequent decade (2006–2015).