SUMMARYEarlier studies demonstrated that oscars, endemic to ion-poor Amazonian waters, are extremely hypoxia tolerant, and exhibit a marked reduction in active unidirectional Na + uptake rate (measured directly) but unchanged net Na + balance during acute exposure to low P O2 , indicating a comparable reduction in whole body Na + efflux rate. However, branchial O 2 transfer factor does not fall. The present study focused on the nature of the efflux reduction in the face of maintained gill O 2 permeability. Direct measurements of 22 Na appearance in the water from bladder-catheterized fish confirmed a rapid 55% fall in unidirectional Na + efflux rate across the gills upon acute exposure to hypoxia (P O2 =10-20 torr; 1 torr=133.3 Pa), which was quickly reversed upon return to normoxia. An exchange diffusion mechanism for Na + is not present, so the reduction in efflux was not directly linked to the reduction in Na + influx. A quickly developing bradycardia occurred during hypoxia. Transepithelial potential, which was sensitive to water [Ca 2+ ], became markedly less negative during hypoxia and was restored upon return to normoxia. Ammonia excretion, net K + loss rates, and 3 H 2 O exchange rates (diffusive water efflux rates) across the gills fell by 55-75% during hypoxia, with recovery during normoxia. Osmotic permeability to water also declined, but the fall (30%) was less than that in diffusive water permeability (70%). In total, these observations indicate a reduction in gill transcellular permeability during hypoxia, a conclusion supported by unchanged branchial efflux rates of the paracellular marker [ 3 H]PEG-4000 during hypoxia and normoxic recovery. At the kidney, glomerular filtration rate, urine flow rate, and tubular Na + reabsorption rate fell in parallel by 70% during hypoxia, facilitating additional reductions in costs and in urinary Na + , K + and ammonia excretion rates. Scanning electron microscopy of the gill epithelium revealed no remodelling at a macro-level, but pronounced changes in surface morphology. Under normoxia, mitochondria-rich cells were exposed only through small apical crypts, and these decreased in number by 47% and in individual area by 65% during 3 h hypoxia. We suggest that a rapid closure of transcellular channels, perhaps effected by pavement cell coverage of the crypts, allows conservation of ions and reduction of ionoregulatory costs without compromise of O 2 exchange capacity during acute hypoxia, a response very different from the traditional osmorespiratory compromise.