This review of water transport measurement in normal human erythrocytes attempts to harmonize discordant results obtained under diverse study conditions with two different techniques: nuclear magnetic resonance (NMR) and radioactive tracer (THO) diffusion. Natural aggregation of red cells into rouleaux appeared to cause most of the variation among results from NMR experiments. The remainder of the discrepancy was attributed to the use of inappropriate mathematical approximations of the two-site exchange equations, differences in blood storage time, and failure to adjust NMR calculations for the nonwater protons. Differences in hematocrit, frequency-magnetic field strength, or NMR pulse technique played no apparent role in the disparity among NMR reports. When these confounding factors were removed, diffusion results obtained by NMR or by influx or bulk diffusion of radioactive tracer agreed within a relatively narrow range of values. These techniques place the mean lifetime of water inside fresh normal human erythrocytes at room temperature (20-25 degrees C) between the extremes of 9.8 and 14 ms, where the uncorrected range was previously 9.8-21.7 ms. This new range of water exchange times corresponds to a range of diffusional permeability between 3.3 and 4.7 x 10(-3) cm/s.