The study investigated serum endosulfan changes resulting from occupational exposure to the pesticide on farms. Eight applicators and 17 non-applicators were tested (serum endosulfan, anthropometry, short exposure questionnaire) before and after the first day of seasonal spraying. Task-based job exposure matrix (JEM) estimates were calculated. Mean baseline serum endosulfan (530 +/- 0.05 microg/L) was high. Increases in post-spraying endosulfan levels (IPSE) were higher in applicators (mean = 60 +/- 90 microg/L) than in non-applicators (mean = 3.5 x 10(- 6)+/- 90.0 microg/L) adjusting for age (beta = 54.0, p = 0.162, R(2) = 0.22). There was a weak positive relationship between IPSE and JEM estimates. IPSE occurred in applicators and non-applicators and were higher in applicators. The validity of the JEM weightings and characterization of other routes of pesticide exposure require further investigation.