The clinical records and autopsy data of 75 patients dying with AIDS were reviewed to determine the frequency of individual diseases diagnosed premortem and postmortem, the significance of pulmonary processes found in the lungs at autopsy, and the clinical and pathologic causes of death. Cytomegalovirus (CMV) infection was identified histologically either premortem or postmortem in 81% of patients. The lungs and adrenal glands were infected most commonly. Only one-half of CMV infections were recognized premortem. Pneumocystis pneumonia and Kaposi sarcoma occurred in 68% and 59% of patients, respectively, but were not unsuspected premortem in any patient. Visceral involvement with Kaposi sarcoma, however, was frequently recognized only at autopsy. While disseminated M. avium-intracellulare infection was common (31% of patients), histologically documented pulmonary disease was uncommon (3% of patients). Cryptococcal infection, diagnosed in 10 patients, was confined to the central nervous system in only 1 patient. Toxoplasma, in contrast, infected the brain of only 6 patients. All 75 patients had one or more disease processes identified in their lungs or pleurae at autopsy. These processes included opportunistic infections in 76% of patients, neoplasms in 37% (Kaposi sarcoma in 36% and lymphoma in 3%), and other processes in 60%. The most prevalent pathogen, CMV was found in pulmonary tissue from 44 patients and caused significant disease in 21 patients. Five patients died due to CMV pneumonia. Pneumocystis carinii was found at autopsy in 24 patients. In spite of treatment, pneumocystis pneumonia was fatal in 11 patients. One patient died with concomitant CMV and pneumocystis pneumonia. Kaposi sarcoma, identified in the lungs of 23 patients, led to death in 5 patients via upper airway obstruction, hemorrhage, or parenchymal destruction. Other fatal pulmonary processes included bacterial pneumonia in 9 patients, idiopathic diffuse alveolar damage in 5, cryptococcosis in 2, and pulmonary hemorrhage in 1. Specific clinical criteria were used to determine the cause of death due to organ system failure. Fifty-one percent of patients died due to respiratory failure; 16% from neurologic disease; 17% from hypotension that was not caused by respiratory, neurologic, or cardiac disease; and 3% from cardiac dysfunction. Thirteen percent of deaths did not meet the clinical criteria defining these 4 categories. This clinical assessment was combined with autopsy data to identify specific diseases as causes of death.(ABSTRACT TRUNCATED AT 400 WORDS)