The potential usefulness and limitations of the double-indicator mean transit time approach for measuring lung water are evaluated from both theoretical and empirical points of view. It is concluded that poor tissue perfusion is the most serious factor that can compromise the reliability of this approach. Replacement of the conventional water isotopes with a thermal signal enhances indicator delivery to ischemic areas but the diffusion of heat is not sufficiently rapid to permit measurements of water in macroscopic collections of fluid which remain unperfused. The frequency of pulmonary vascular obstruction in patients with pulmonary edema related to lung injury suggests that interpretation of transit time data will be complicated by uncertainties concerning perfusion. Thermal-dye measurements of lung water may prove more helpful in situations where pulmonary blood flow remains relatively uniform.