To detect cytomegalovirus-associated interstitial pneumonia (CMV-IP) in recipients of BMT in its earliest stage, five CMV methods were assessed for their usefulness using bronchoalveolar lavage fluid as the test specimen. Of the 43 cases enrolled in the study, PCR was positive in 12 cases, shell vial in eight, culture in eight and cytology in three. There were no positive cases in in situ hybridization. Based on this result, the 43 cases were classified into four groups: Group 1, three cases: positive in PCR, shell vial and cytology; Group 2, five cases: positive in PCR and shell vial; Group 3, four cases: positive only in PCR; and Group 4, 31 cases: negative in all CMV tests. Cases in Group 1 were judged as having the highest risk of overt CMV-IP. They were successfully treated with a combination of ganciclovir and immunoglobulin. Group 2 was diagnosed as having active CMV infection and ganciclovir monotherapy was effective for these patients. Groups 3 and 4 were not given anti-CMV therapy, but they were free from CMV-related manifestations throughout the study. The sensitivity and specificity of each survey method for the detection of Groups 1 and 2 were 1.0 and 0.89 in PCR, 1.0 and 1.0 in shell vial, 0.88 and 1.0 in culture, and 0.38 and 1.0 in cytology. Similarly, the positive and negative predictive values were 0.67 and 1.0 in PCR, 1.0 and 1.0 in shell vial, 1.0 and 0.97 in culture, and 1.0 and 0.88 in cytology. Thus, CMV survey on bronchoalveolar fluid was thought to be useful in detecting post BMT CMV-IP in its earliest stage.