To determine if the use of aerosolized pentamidine prophylaxis decreases the clinical severity or the sensitivity of diagnostic tests for Pneumocystis carinii pneumonia (PCP), we conducted a retrospective matched cohort comparison study of patients admitted to San Francisco General Hospital with PCP from August 1, 1989, to June 30, 1990. Patients who had received pentamidine prophylaxis during at least the 2 months prior to the diagnosis of PCP were matched with patients who had not received the drug. Matching was based on the number of prior episodes of PCP, sex, age, and risk factors for human immunodeficiency virus infection. As markers of clinical severity, we chose alveolar-arterial oxygen difference, serum lactate dehydrogenase levels, outpatient versus inpatient treatment, length of hospitalization, length of intravenous anti-pneumocystis treatment, development of respiratory failure, in-hospital mortality, and chest radiographic appearance. Although, of the 27 matched pairs identified, significantly fewer of the pentamidine cohort were treated as inpatients, and significantly more of this cohort had upper lobe dominant disease on chest radiograph, we found no other significant differences between markers of clinical severity for the two cohorts. In addition, we found no significant differences in the rate of sputum or bronchoalveolar lavage positivity for P. carinii between the two cohorts. We conclude that, although hospitalization is less common in patients with a history of prophylactic pentamidine use, aerosolized pentamidine prophylaxis does not decrease the clinical severity or the sensitivities of sputum induction or bronchoalveolar lavage as diagnostic tests for PCP.