The relationship between serologically confirmed cases of leptospirosis and the amount of rainfall in greater Florianôpolis, in southern Brazil, was studied retrospectively for the period 1991 to 1996. Maximum daily and total monthly rainfall for present and previous months were statistically significant predictors of the number of cases of leptospirosis in Poisson regression models. However, daily maximum rainfall data showed a much better model fit than total monthly rainfall. For each millimeter increase in maximum daily rainfall for the month above the average for the period studied, there was an increase of 0.55% in the number of leptospirosis cases relative to the period average. For the past month's daily maximum, this increase was 0.21%. Maximum daily rainfall during the month is a sensitive indicator of large amounts of rain falling in a brief period of time. This may cause flash floods and, thus, disseminate pathogenic Leptospira among the human population, particularly in densely populated areas with large rodent populations and without adequate drainage. This situation is typical for the association of tropical rainstorms and the spread of leptospirosis among slum dwellers.