We evaluated the accuracy, reproducibility, and linearity of lung water concentration (LWC) measurements using positron emission tomography (PET) in anesthetized supine dogs. First, we evaluated whether errors in attenuation correction, which might occur during the development of pulmonary edema, significantly affect LWC and pulmonary blood flow (PBF) measurements obtained with PET. In 10 animals, PET scans of LWC and PBF were obtained before and after oleic acid induced lung injury. Changes in LWC and PBF after injury were underestimated by 12%-25% if the postinjury LWC scan was reconstructed using attenuation data obtained prior to injury, before lung density had changed, instead of attenuation data obtained after lung injury. Next, in four other dogs, we did not administer oleic acid, but instead instilled progressive amounts of autologous plasma into the left caudal lobe and obtained LWC measurements after each. Changes in LWC were linearly related to the amount of plasma instilled (r = 0.99). The average coefficient of variation for LWC in the control right lobe was 4 +/- 2% and the average percent change between measurements was 1.5% +/- 5.8%. The correlation of regional LWC determined gravimetrically after the last scan, when corrected for differences in regional lung density, with LWC determined by PET, was excellent (r = 0.92). We conclude that when the correct attenuation scan is used for emission scan reconstruction. PET measurements of LWC are accurate, linear, and reproducible.