The accuracy of clinical diagnoses was determined and compared between the two periods of time 1977 to 1978 and 1987 to 1988 based on the analysis of 3,042 autopsies at Huddinge University Hospital. The discrepancy rates were calculated by counting the number of missed or incompletely diagnosed major diseases and their complications. Moreover, sensitivity, specificity, and clinical accuracy for positive and negative diagnoses also were calculated for all cases of acute myocardial infarction and malignant tumors. The autopsy rate decreased from 80% to 39%. The autopsy successfully addressed the clinical questions in 97% of the cases. The selection of the cases possibly could explain the significant 5% increase in the proportion of clinically undetected major underlying diseases. The discrepancy rate was higher among the older patients. There were no significant changes in the diagnosis of cardiovascular diseases apart from thrombosis of the mesenteric artery, which more than doubled. The proportion of infectious diseases increased significantly from 27% to 32%. The number of cases with clinically missed tuberculosis was twice as high in the 1987 to 1988 period as in the 1977 to 1978 period and there was a marked increase in fungal and viral infections. There was no significant change in the clinical diagnosis of malignant tumors between the two periods. Approximately 15% of all major cancers were not diagnosed before autopsy; half of these tumors were lethal. We conclude that the role of the autopsy has not diminished in spite of advanced diagnostic methods and it remains an effective tool in the assessment of medical care.