We analyzed population-based childhood cancer incidence rates throughout California in relation to agricultural pesticide use. During 1988-1994, a total of 7,143 cases of invasive cancer were diagnosed among children under 15 years of age in California. Building on the availability of high-quality population-based cancer incidence information from the California Cancer Registry, population data from the U.S. Census, and uniquely comprehensive agricultural pesticide use information from California's Department of Pesticide Regulation, we used a geographic information system to assign summary population, exposure, and outcome attributes at the block group level. We used Poisson regression to estimate rate ratios (RRs) by pesticide use density adjusted for race/ethnicity, age, and sex for all types of childhood cancer combined and separately for the leukemias and central nervous system cancers. We generally found no association between pesticide use density and childhood cancer incidence rates. The RR for all cancers was 0.95 [95% confidence interval (CI), 0.80-1.13] for block groups in the 90th percentile and above for use of pesticides classified as probable carcinogens, compared to the block groups with use of < 1 lb/mi(2). The RRs were similar for leukemia and central nervous system cancers. Childhood leukemia rates were significantly elevated (RR = 1.48; 95% CI, 1.03-2.13) in block groups with the highest use of propargite, although we saw no dose-response trend with increasing exposure categories. Results were unchanged by further adjustment for socioeconomic status and urbanization.