We have previously reported a method for measuring regional pulmonary blood flow (PBF) in experimental animals using O-15-water and PET. The method requires withdrawing blood from the pulmonary artery during the PET scan, so that the input function can be estimated for the one-compartment model used to analyze the data. The purpose of the present study was to modify and validate this technique for a more general use in humans. Methods: PBF was measured after injections of O-15-water in 15 normal subjects and in five patients with reduced cardiac output. In ten of the normal subjects, PBF was also measured after the injection of Ga-68-albumin macroaggregates (MAA). In the five other normal subjects and in the cardiomyopathy patients, PBF was measured twice after two separate O-15-water administrations. The input function was estimated from a region of interest (ROI) over the right ventricle (RV), with corrections when necessary, for time delays between RV and lung tissue. Results: The mean value for PBF in the normal subjects was 121 +/- 32 ml/min/100 ml lung, and was 57 +/- 33 ml/min/100 ml lung in the patients with cardiomyopathy. The correlation between PBF measured with O-15-water and PBF measured with Ga-68-MAA was r = 0.96. There was no significant difference in the mean value for PBF or the ventral-dorsal distribution of PBF when sequential measurements were made in the same individual. PBF increased in general in the ventral-dorsal direction in these supine subjects, although PBF was more evenly distributed in the cardiomyopathy patients. Conclusion: Measurement of regional PBF with O-15-water and PET appears to be a valid, noninvasive approach for evaluating the pulmonary perfusion pattern of humans.