Objective: to describe spatial patterns of intentional homicides in João Pessoa, Paraíba, Brazil, 2011-2016. Methods: this was an ecological study using census tracts as units of analysis; the study used data on intentional homicides of victims living in João Pessoa obtained by cross-checking Mortality Information System (SIM) information with Public Security information; mortality coefficients were calculated for the whole period and its triennia and were later smoothed using the local empirical Bayesian method; spatial autocorrelation techniques were applied to the smoothed coefficients which were using Moran's spatial statistics. Results: significant spatial autocorrelation was detected for the period as a whole (I=0.679, p=0.01), the 1st triennium (I=0.508, p=0.01), and the 2nd triennium (I=0.572, p=0.01); areas of greater risk were identified distributed among the western, northwestern, southeastern, and far south zones of the city, mainly in regions with low socioeconomic conditions; census tracts with low coefficients were located in areas of medium/high socioeconomic status. Conclusion: homicides in João Pessoa were high in poor districts and these are priority areas for intervention.