From January 1 to December 31 of 1975, all patients admitted to Salvador's Infectious Disease Hospital were serologically tested for leptospirosis if their clinical and laboratory findings were compatible with this diagnosis. Of the 888 patients tested, 133 were positive. Among patients with an initial clinical impression of leptospirosis, the diagnosis was confirmed serologically in 36% and for the 133 leptospirosis patients, only 67 (50.4%) were initially suspected. Epidemiological aspects investigated covered: age, sex, place of residence and occupation for all cases and for fatal cases; distribution of cases by month of occurrence and by rainfall in that month; distribution of serotypes by month; and probable source of infection with emphasis of the occurrence and type of contact with dogs. For each case residing in Salvador, a control of the same age and sex was interviewed in the immediate neighborhood. Sewage, rats, water, dogs, mud and garbage were identified as the most likely source of infection in that order. Marked seasonal fluctuations were apparent with peak incidence during the months of maximum precipitation. In two-thirds of cases reporting contact with a dog, tests on the animal revealed the same serotype. The most frequent serotypes were: icterohemorrhagiae, autumnalis, castellonis, grippotyphosa, hebdomadis and bataviae.