In the traditional osmorespiratory compromise, fish increase their effective gill permeability to O2 during exercise or hypoxia, and in consequence suffer unfavorable ionic and osmotic fluxes. However oscars, which live in the frequently hypoxic ion-poor waters of the Amazon, actually decrease ionic fluxes across the gills during acute hypoxia without changing gill paracellular permeability, and exhibit rapid paving over of the mitochondrial-rich cells (MRCs). But what happens during prolonged exercise? Gill paracellular permeability, ionic fluxes, and gill morphology were examined in juvenile oscars at rest and during aerobic swimming. Initial validation tests with urinary catheterized fish quantified drinking, glomerular filtration, and urinary flow rates, and confirmed that measurements of gill paracellular permeability as [(3)H]PEG-4000 clearances were the same in efflux and influx directions, but far lower than previously measured in comparably sized trout. Although the oscars achieved a very similar proportional increase (90 %) in oxygen consumption (MO2) to trout during steady-state swimming at 1.2 body lengths sec(-1), there was no increase in gill paracellular permeability, in contrast to trout. However, oscars did exhibit increased unidirectional Na(+) efflux and net K(+) rates during exercise, but no change in drinking rate. There were no changes in MRC numbers or exposure, or other alterations in gill morphology during exercise. A substantial interlamellar cell mass (ILCM) that covered the lamellae to a depth of 30 % was unchanged by 4 h of swimming activity. We conclude that a low branchial paracellular permeability which can be dissociated from changes in O2 flux, as well as the presence of the ILCM, may be adaptive in limiting ionoregulatory costs for a species endemic to ion-poor, frequently hypoxic waters.