During 1984-88 a population-based case-control study was carried out in The Netherlands in collaboration with the International Agency for Research on Cancer in order to further explore the possible relationship between diet and exocrine pancreatic carcinoma. Past habitual dietary intake was assessed in 164 cases and 480 controls. This is the first series of population-based studies of pancreas cancer to perform a comprehensive assessment of diet. The results of logistic regression analysis, controlled for age, gender and total cigarette consumption, suggested a positive association with past habitual intake of energy in directly interviewed patients (OR highest versus lowest quintile 3.35; 95% confidence interval 1.51-7.18). No indication of an effect of Quetelet index was found. When maximum-weight-ever-achieved was used, nonsignificantly reduced risks were observed for the upper 4 quintiles of "maximum" Quetelet index. After controlling for age, gender, response status and total cigarette consumption, a positive relationship with habitual past intake of total carbohydrates was observed (OR 2.40; 95% confidence interval 1.44-3.99). The effect decreased considerably when the analysis was restricted to subjects who reported stable long-term dietary intake. After controlling for age, gender, total cigarette consumption and response status, a significant, positive association between pancreatic cancer and past habitual intake of simple sugars was found (OR 1.95; 95% confidence interval 1.24-3.07). Among those who reported stable long-term dietary intake, no effect was seen. The positive effect of simple sugars was present in males only (OR 2.15; 95% confidence interval 1.18-3.93) and was smaller in men who reported long-term stable dietary intake. Our findings suggest that the development of exocrine pancreatic carcinoma is positively related to past habitual intake of total energy, total carbohydrates and simple sugars, whereas no relationship with body-mass index was observed.