This study describes the mortality experience in a cohort of 23,401 farmers, residing in southern Piedmont, Italy, and licensed to use pesticides. From 1970 to 1986 the cohort included 340,794 person-years and 2683 deaths were observed. A strong attenuation of the death risk was found due to the healthy worker effect (seen as an active role in the application for the license by the members of the cohort) and due to the limited comparability of the cohort with respect to the reference population. The standardized mortality ratios (SMRs) were remarkably < 100 for all causes (SMR = 59; 95% confidence interval = 57-61) and for all tumors (SMR = 60; 95% CI 55-64), but they increased with the increasing duration of the follow-up. A risk increase was observed with respect to melanomas and eye tumors in the entire cohort and lymphoma and tumors of the connective tissue in the subcohort of subjects living in villages with mainly arable land.