It would be quite reasonable for us to expect the progress made in diagnostic technology to be accompanied by a parallel improvement in diagnostic accuracy. In reality, however, the frequency of misdiagnoses remains the same, despite the fast progress which has been made by medical technology in the last 30 years. Autopsy is the best source of information on diagnostic accuracy. According to one hypothesis, an increase in the number of autopsies performed and the follow-up on them could reduce the number of diagnostic mistakes. In recent times, however, the number of autopsies in comparison with the registered number of deaths has been declining steeply. We studied the autopsy reports for 1997, kept at the archive of the Institute for Forensic Medicine. We only took into account the deaths which occurred within 24 h of admittance to the emergency wards of the Ljubljana University Hospital, including those patients who died subsequently as a consequence of accident or injury. We also included cases of sudden deaths which occurred during operating or within 24 h after it. Following selection, we analyzed 444 out of the total of 921 autopsy reports, for each of which we carried out a comparison between the postmortem diagnosis and the clinical diagnosis, contained in the medical report on the death and the causes of death, which is modeled on WHO recommendations, i.e., the International Classification of Diseases (ICD), and in the medical documents, if any were submitted. Data are entered in these by using the ABC system where: A) direct cause of death, B) are circumstances that influenced the occurrence of death, and C) is original cause of death. The findings were then organized into five groups, depending on the degree to which the clinical diagnosis agreed with the postmortem diagnosis. The first group is comprised of the cases where the clinical and postmortem diagnoses agree completely; the second group is comprised of the cases of partial disagreement on the direct cause of death; the third group is comprised of the cases of disagreement on the original disease; the fourth group, of complete disagreement between the clinical and postmortem diagnoses. The fifth group is comprised of those cases where, under the ABC standards on the classification of diseases, injuries and causes of deaths as specified by the WHO, the documentation was incomplete. A complete agreement between the diagnoses was established in 48.87% of cases; partial disagreement in 22.74%; and total disagreement in 13.5%. 9.68% of cases were classified as falling into group 5. For the three diseases that are among the most common causes of death, we established the percentage of agreement, the percentage of overdiagnosis and the percentage of underdiagnosis. The most frequently underdiagnosed disease (in 61% of cases) was pulmonary thromboembolia; in 15% a thromboembolia was confirmed in autopsy. In 24%, a myocardial infarction was not diagnosed clinically and in 60% the clinical diagnosis of a myocardial infarction was confirmed in autopsy. In 33% a heart failure was not diagnosed during the clinical stages but only in autopsy, in 66% the clinical diagnosis of a heart failure was confirmed in autopsy.