Two series of data from the California State Department of Health and one from Children's Hospital of Los Angeles were analyzed to determine whether an association exists between influenza epidemics and incidence of leukemia in children born following such outbreaks. Of leukemic children born in areas for which information on past influenza activity was available, the population-based Alameda County Cancer Registry recorded 89 cases during 1960-1969, the California Tumor Registry recorded 653 cases during 1950-1970, and Children's Hospital recorded 575 cases during 1957-1972. Flu epidemics were identified and flu cohorts constructed such that leukemic children who could have been in utero during a flu epidemic constituted a flu cohort. Based on the total number of leukemia cases reported and the proportion of at-risk weeks contributed by the flu cohort, expected numbers of leukemia cases from the flu cohort were computed for each age. In each series, there was an excess of leukemia cases in the flu cohort, contributed primarily by the 0-4 age group. Trimester analysis indicated that the greatest excess occurred in children who were in the first trimester of gestation during a flu epidemic. Incidence data from the Cancer Incidence System of the San Francisco Bay Area Resource for Cancer Epidemiology and the Third National Cancer Survey show a relative risk of 3.4 for this group. Due to sources of misclassification in this study, such as analysis of flu cohorts rather than individuals whose mothers actually had influenza during pregnancy, the results tend to be diluted. If influenza is a causative factor, the actual increase in leukemia risk is likely to be much greater than this analysis indicates.