Published values for sperm membrane water permeability (L(p)) obtained using a time-to-lysis methodology have produced anomalous results when used to model optimal cooling rates for cryopreservation of spermatozoa. As the lysis method is dependent on potentially questionable assumptions, we describe an alternative method for measuring sperm L(p). Spermatozoa were exposed to hypo- and hyperosmotic conditions using a stopped-flow apparatus and the time course of resulting volume changes was measured using concentration-dependent self-quenching of the entrapped fluorophore, carboxyfluorescein (CF). L(p) was measured for boar, rabbit, and ram spermatozoa using a range of osmotic stresses (+/-50-100 mOsm). Values for exosmotic and endosmotic flow showed no evidence of rectification. Mean L(p) values were 0.84 microm/min/atm (boar), 0.28 microm/min/atm (rabbit), and 2.79 microm/min/atm (ram). These values are lower than the lysis method estimates, with the ram value reduced by approximately two-thirds using the current methodology. The value for boar spermatozoa showed good agreement with published values obtained using an electronic cell-sizing technique. Substitution of the revised values for L(p) into the model for optimal cooling rates brings the calculated optimal rate closer to the lower empirically observed value but does not fully account for the previously reported discrepancies. Copyright 2000 Academic Press.