We undertook a population-based cohort study in Pizarra (Spain). Anthropometric and nutritional variables were recorded for 613 persons. The type of fat used was determined by measurement of the fatty acids contained in cooking oil. Serum fatty acid was used as a biological marker of the type of fat consumed. Obesity incidence in persons who were not obese at baseline was greater in those who consumed sunflower oil (Group 1: 41.5 (95% CI, 25.4-67.8) cases per 1000 person-years) than in those who consumed olive oil or a mixture of oils (Group 2: 17.3 (95% CI, 11.6-25.8) cases per 1000 person-years). The risk of developing obesity over 6 years, adjusted for age, sex, physical activity, smoking, instruction level, energy intake and baseline BMI, was 2.3 (95% CI, 1.06-5.02) in group 1 compared with that in group 2. The increase in the prevalence of obesity in the free-living population is associated with the type of fatty acids in the diet.