Obesity may be associated with increased risk of prostate cancer (PCa). According to one hypothesis, obesity could lower the risk of non-aggressive tumours, while simultaneously increasing the risk of aggressive cancer. Furthermore, central adiposity may be independently associated with PCa risk; it is also associated with diabetes, which itself may influence risk of PCa. We studied the associations between height, body composition, and fat distribution, diabetes prevalence and risk of total, aggressive, and non-aggressive PCa in 10 564 initially cancer-free men (aged 45–73 years) of the population-based Malmö Diet and Cancer cohort. Anthropometric and body composition measurements, including bioelectrical impedance for estimation of fat mass, were performed by study nurses. Diabetes prevalence was self-reported. Cancer cases and clinical characteristics were ascertained through national and regional registry data. Dietary and other background data were obtained through a modified diet history method and an extensive questionnaire. During a mean follow-up of 11.0 years, 817 incidental PCa cases were diagnosed. Of these, 281 were classified as aggressive. There were 202 cases occurring before 65 years of age. Height was positively associated with total and non-aggressive PCa risk. Waist–hip ratio (WHR), a measure of central adiposity, was positively associated with PCa before age 65, and less strongly, with total PCa. This association was independent of body mass index (BMI) and other potential confounders. General adiposity, expressed as BMI or body fat percentage, and prevalent diabetes were not associated with PCa risk. In this study, WHR and body height were stronger PCa predictors than general adiposity.