In doing a case-control study of cancer of the exocrine pancreas, a set of criteria was developed to classify cancer of the exocrine pancreas by pathological and diagnostic data as to the probability that the tumor arose in the pancreas. We applied these criteria to the records of 125 consecutive patients between the ages of 40-79 years who were identified as having adenocarcinomas of the exocrine pancreas from a population-based cancer registry. We found that histological confirmation by a cancer registry does not guarantee the site of origin is the pancreas. Of the histologically confirmed cases meeting our study criteria, 28.7% may not have arisen in the pancreas. Forty-seven per cent of non-histologically confirmed cases, which most studies exclude, were almost certainly of pancreatic origin based on computerized tomogram (CT) scan and/or surgical palpation and subsequent clinical course. These figures may underestimate the full extent of diagnostic misclassification, especially in older age groups. We conclude that epidemiological studies of cancer of the exocrine pancreas need more uniform criteria for defining cases, and these criteria should take into account uncertainty as to site of origin. Sufficient detail of our scheme is presented so it can be applied in other studies.