The objective of this research was to analyze the relationship between socioeconomic statistics and homicide mortality rates in the city of São Paulo between 1988 and 1994. City districts were grouped into five geographic areas. A socioeconomic indicator (ISE) was constructed with census information combining average income of the family head, illiteracy rate for the population over 5 yr of age, average number of rooms and number of persons per household. The higher the score, the better the socioeconomic situation (possible values: 4 to 384). Deaths from homicide were grouped by residential areas, and the rates for these areas were calculated. The association between homicide rates and the socioeconomic situation was analyzed with Spearman correlation coefficients. Average ISE scores varied from 291 in the Central area (homicide rate = 27.96 deaths per 100000) to 119.9 in the East area (homicide rate = 40.38). The Spearman coefficient between ISEs and homicide rates was -0.98 (p < 0.05). Due to the heterogeneity inside the areas, the median ISE is a better indicator of the socioeconomic conditions yielding an rs = -1.0. Almost half the population resides in areas with the highest risk of homicide mortality (East and South). Taking the Central area as a reference, we found risks of 1.36 in the Western, 1.37 in the Northern, 1.44 in the Eastern and 2.67 in the Southern areas.