The aims of the study were to assess the agreement between data from personal interviews and medical records on selected medical conditions, to evaluate the validity of each source of information, and to estimate the prevalence of these conditions in pancreatobiliary pathology. Between 1992 and 1995, 602 patients with pancreatobiliary diseases were included in the PANKRAS II Study. Information from interview and medical records was available in 88% of cases. The agreement between both data sources was poor only for cholecystitis and mental disorders. The validity of both sources of information was high or acceptable for all conditions except for cholecystitis. Prevalence of past medical conditions was similar to that reported previously and varied according to diagnosis, gender, age, educational level, and hospital of admission. Research aimed at assessing the prevalence of less common diseases should consider subject- and health-related factors in collecting information.