Epidemiologic studies of farmers have linked pesticides with certain cancers. Information on exposures from many of these studies was obtained by interview of farmers or their next-of-kin. The reliability and validity of data on pesticide use obtained by recall, often years after the event, have been questioned. Pesticide use, however, is an integral component in most agricultural operations, and the farmers' knowledge and recall of chemicals used may be better than for many other occupations. Contrary to general belief, many farmers typically use only a few pesticides during their lifetimes and make only a few applications per year. Data from U.S. Department of Agriculture surveys indicate that herbicides are applied to wheat, corn, soybeans, and cotton and that application of insecticides to corn averages two or fewer times per year. In epidemiologic studies at the National Cancer Institute, the proportion of farmers ever reporting lifetime use of five or more different chemicals was 7% for insecticides and 20% for herbicides. Surrogate respondents have often been used in epidemiologic studies of cancer; they are able to recall pesticide use with less detail than the farmers themselves. The pesticides reported by surrogates were the same as reported by subjects themselves, but with less frequency. Comparison of reporting by cases and controls provided no evidence of case-response (differential) bias; thus, inaccurate recall of pesticide use by subjects or surrogates would tend to diminish risk estimates and dilute exposure-response gradients.