BACKGROUND AND AIM: Epidemiological studies have reported an increased risk of Parkinson's disease (PD) in farmers exposed to pesticides, but no clear conclusion can be drawn on the type of pesticide and duration of use associated with an effect. In the French agricultural cohort AGRICAN, we assessed associations between PD and pesticide use according to the types of livestock and crops grown, including exposure to some active ingredients with duration of use. METHODS: Self-reported PD and history of lifetime exposure to 13 crops and 5 types of animals and pesticide use were collected at enrolment (2005-07) among 181 842 participants. Exposure to selected active ingredients and duration of use lifelong were assessed with the crop-exposure matrix PESTIMAT. Associations between pesticide use and PD were estimated by logistic regression according to crops and livestock, adjusted for sex, age, educational level, smoking status and alcohol consumption. RESULTS: PD was reported by 1732 subjects (1.2%) at enrolment in the cohort. Pesticide use lifelong was associated with an increased risk of PD in all types of activities [odds ratio (OR) = 1.31 (cattle) to 1.79 (peas), P < 0.05]. Rotenone, diquat, paraquat and several dithiocarbamates were associated with an increased risk of PD [OR = 1.31 (cuprobam) to 1.57 (rotenone)], especially in farmers with the longest exposure. CONCLUSIONS: Our work suggests that the risk of PD is increased in farmers exposed to pesticides on several French crops and livestock, and supports additional evidence of an association of PD with dithiocarbamate fungicides, rotenone and the herbicides diquat and paraquat.