The purpose of the study was to determine (a) the frequency and cause of mortality in patients with chronic pancreatitis; (b) the cumulative survival rates corrected by comparison of patients with a matched French population; and (c) the factors associated with mortality by a unidimensional and multidimensional analysis. The study population consisted of 240 patients (men = 208, women = 32; alcoholic = 210, nonalcoholic = 30) followed for a mean time of 8.7 yr. The status of the patients (dead or alive) was recorded in February 1987. Mean age at onset of chronic pancreatitis was 41.5 yr. Fifty-seven patients died. Mean age at time of death was 52.3 yr. "Overmortality" after 20 yr of course was 35.8% in comparison with a matched French population (p less than 0.0001). Chronic pancreatitis was the direct cause of death for only 19.3% of patients. The main causes of death have been alcoholic hepatopathy (n = 10), cancer (n = 9), postoperative mortality (n = 8). Unidimensional analysis of mortality rates showed that male sex (p less than 0.03), surgery (p less than 0.007), hepatopathy (p less than 0.01), diabetes mellitus (p less than 0.02), and absence of attack of acute pancreatitis (p less than 0.02) were associated with mortality. Multidimensional analysis showed that the following variables were linked with mortality: in a first model including the totality of the study population: surgery (p less than 0.006), hepatopathy (p less than 0.008), no attack of acute pancreatitis (p less than 0.03), male sex (p less than 0.03); in a second model excluding cirrhosis: surgery (p less than 0.001), male sex (p less than 0.06), diabetes mellitus (p less than 0.09). Nevertheless, surgery did not seem to interfere with long-term mortality. The lower mortality of patients with attacks of acute pancreatitis suggests a favorable influence for alcohol abstinence.