Using data collected of a large population-based cohort study, we studied the association between anthropometric factors and the risk of pancreatic cancer. Furthermore, we investigated whether these associations differ among microscopically confirmed pancreatic cancer (MCPC) cases and non-MCPC (NMCPC) cases. The Netherlands Cohort Study on Diet and Cancer started in 1986 (120,852 men and women) and uses the case-cohort methodology. After 13.3 years of follow-up, 446 pancreatic cancer cases (of which 65% was microscopically confirmed) and 4,774 subcohort members were available for analysis. The multivariable incidence rate ratio of MCPC of men was 1.10 per increment of 1 kg.m(-2) (95% confidence interval, 1.04-1.18). Women had a rate ratio of MCPC of 1.08 (95% confidence interval, 1.03-1.13). Obese men [body mass index (BMI) >/=30 kg.m(-2)] had a 2.6-fold increased risk of MCPC compared with men with BMI 23 to 25 kg.m(-2). For women, this increase in risk was 1.7-fold. Change in BMI between age 20 years and baseline was also associated with MCPC in both men and women. In men and women, none of these associations were observed for NMCPC, with the exception of the increased risk for pancreatic cancer in obese men. We observed statistically significant associations between both BMI, gain in BMI, and pancreatic cancer risk. These associations are observed only in MCPC and not in NMCPC. If MCPC and NMCPC had been considered as one group, the reported associations would not have been detected. These findings stress the need to evaluate heterogeneity among pancreatic cancer cases in etiologic studies. (Cancer Epidemiol Biomarkers Prev 2007;16(7):OF1-6).