The water content of lung, brain, and muscle tissue was measured by nuclear magnetic resonance (NMR) and compared with gravimetric determinations. The NMR signal intensity of water was measured by a single 90 degree pulse and by a spin-echo sequence. The absolute water content was determined by the difference in the sample's weight before and after desiccation. The NMR detectable water in each tissue was expressed as a percentage of the signal intensity for an equal weight of distilled water. Using the single pulse measurement, 67% of the gravimetrically-measured water was detected in collapsed lung samples (consisting of about 47% retained air), in contrast to 96% for brain and 98% for muscle. For degassed lung samples, the NMR detectability of water increased to 87% with the single pulse measurement and to 90% with the spin-echo measurement, but the values remained significantly less than those of brain or muscle. Factors that caused the NMR signal loss of 33% in collapsed lung samples were: air-tissue interfaces (20%), microscopic field inhomogeneity (3%), and a water component with an extremely short magnetization decay time constant (10%).