Fine needle aspiration cytologic specimens from pancreatic lesions in 77 patients were blindly reviewed. The patients were divided into two subgroups: group A included 19 patients without malignancy of the pancreas, and group B included 58 patients with confirmed malignancy (primary or secondary) of the pancreas. The original cytologic diagnoses in both groups were plotted against the respective reevaluated cytologic diagnoses. In group A there were no false-positive cytologic diagnoses, neither originally nor on reevaluation. In group B, among nine cases originally diagnosed as "benign cells," there were two cases diagnosed as malignant on reevaluation, while in the remaining seven cases malignant cells could not be found on reevaluation. In 11 cases in group B the cell material was considered insufficient for cytologic diagnosis, both originally and on reevaluation. In 34 of the 58 cases (58.6%) in group B a conclusive cytologic diagnosis of malignancy was made both in the original cytologic reports and on reevaluation. In this series, sampling errors accounted for most false-negative cytologic diagnoses. The specificity in our series was 100%. The demonstrated sensitivity was 59% initially and 62% on reevaluation.