Identification of dietary and lifestyle variables associated with the development of Parkinson's disease (PD) may offer pathogenetic clues and prevention opportunities. In a population-based prospective cohort study, 26,173 participants in the EPIC-Greece cohort had sociodemographic, anthropometric, medical, dietary and lifestyle variables ascertained at enrolment and periodically reassessed with follow-up contacts. Based on these data, subjects were screened as possible PD cases if they (1) reported either a medical diagnosis of PD or use of anti-PD drugs and (2) did not report preceding causes of secondary parkinsonism. For diagnostic validation, possible incident PD cases were assessed by a focused 3-item telephone questionnaire. Cox proportional hazards regression was used to evaluate associations between potential predictors and incident PD. The main multivariate model included gender, age, marital status, schooling years, farming occupation, smoking status, caffeinated coffee, body mass index, physical activity and energy intake. Additional models included all above variables plus one dietary item at a time. Incidence rate adjusted to the European population was 16.9 per 100,000 person-years. In multivariate models, incident PD exhibited strong positive association with consumption of milk, but not cheese or yoghurt. This finding may help narrow down the search for potential dairy product components with a facilitatory role in PD. Concerning other dietary components, inverse association was found between polyunsaturated fat intake and incident PD. Also, inverse association was found with tobacco smoking, in line with previous studies, but not with caffeine.