This study was designed to assess the efficacy of using oral washes (OWs) to diagnose Pneumocystis carinii pneumonia (PCP) in patients with a low parasite burden and to detect cases of subclinical infection. A total of 104 paired induced sputum (IS) samples and OWs from 104 HIV-seropositive patients and 32 OWs from immunocompetent healthy controls were studied. All of the control samples were negative. Fifty-two IS specimens were positive for Pneumocystis carinii, and 26 of these cases were also detected in the OWs using conventional stain or polymerase chain reaction. Twenty-four of the PCP cases had a high or a moderate parasite load and 28 had a low parasite load; among them, Pneumocystis carinii was detected in the OWs of 15 and 11 cases, respectively. Fifteen of the 104 IS samples studied belonged to patients who were asymptomatic carriers or who had a subclinical infection, and Pneumocystis carinii was detected in the OWs of 4 of these cases. The parasite was not detected in 37 IS samples and in 74 OWs. The results of this study indicate that in patients with a low pulmonary parasite burden, the number of organisms reaching the oral cavity is insufficient for reliable detection in OWs. Thus, OWs are less useful than IS samples for detecting Pneumocystis carinii in cases of pneumonia in which a low parasite burden and/or subclinical infection are present.