The authors utilized a recently developed DNA probe technique to obtain quantitative data on occurrence of Vibrio cholerae in samples collected monthly from 12 environmental sites in Lima, Peru, from November 1993 through March 1995. Peak V. cholerae counts ranged from 10(2)/ml to 10(5)/ml, with the highest counts in sewage-contaminated areas and irrigation water. With our methodology, no V. cholerae cases were detected at any site during the winter months of July through October. Counts were detectable in the environment before onset of cholera in the community, with counts at "cleaner" sites upriver correlating significantly with occurrence of community disease 2 and 3 months later. In sites with heavy sewage contamination, V. cholerae could still be detected before the onset of cases in the community; however, in contrast to upriver sites, counts at these latter sites correlated most closely with the number of concurrently occurring cholera cases. These data support a model of cholera seasonality in which initial increases in number of V. cholerae in the environment (possibly triggered by temperature) are followed by onset of illness in the community, with these human cases further amplifying the organism as the epidemic cycle proceeds.