1. The temperature-dependence of the uptake of 24Na and 42K into dog red cells between 38 and 4 degrees C has been investigated. The effects on the cation fluxes of partial dehydration of the cells in hyperosmolar sucrose (50-125 mM) have also been studied. 2. A Hamilton gas-tight syringe was used to pipette accurately reproducible volumes of packed cells which contained in addition to 24Na or 42K either [131I]albumin or [51Cr]EDTA as extracellular markers. 3. At 38 degrees C Na flux (m-equiv/l. isosmolar cell volume. hr) increased from 2-8 +/- 0-1 (n = 8) in cells of normal volume to 226 +/- 8 (n = 8) when the cells were shrunken by 27-4 +/- 0-6% (n = 8) in media containing sucrose (100 mM). K influx remained relatively constant under these conditions. 4. The exchange of 24Na in shrunken cells followed a single exponential time course but about 9% of the intracellular Na apparently did not exchange with 24Na in the bathing medium. 5. The steady-state influx of Na in cells of normal volume was maximal at about 22 degrees C. The temperature dependence of the Na fluxes in shrunken cells was described by an Arrhenius relationship with a change in slope at about 22 degrees C. 6. The K influx in cells of normal volume decreased as the temperature was lowered from 38 degrees C, to about 12 degrees C, at which temperature the flux was at a well defined minimum. Above 12 degrees C, cell shrinkage had hardly any effect on K influx, but below 12 degrees C the influx in shrunken cells was significantly less than in cells of normal volume. 7. The selective increase in Na flux induced by cell shrinkage results from a Na:Na exchange process which cannot be explained in terms of Ussing's (1947) model of carrier-mediated exchange diffusion. 8. The lack of coupling between the effects of temperature and cell volume on the fluxes of Na and K indicates that localized structural changes of lipid-protein complexes specific for Na or K are responsible for the cation transport characteristics of dog red cells, and that phase transitions in the lipids of the cell membrane are unlikely to account for the temperature dependence of the fluxes.