1. In rats anaesthetized with Saffan, renal function was monitored from the left kidney from the 5th minute of spontaneous breathing of 12% O2 for two 20 min periods and during air breathing before, between and after the hypoxic periods. Two groups of animals (I and II) were used, each group comprising two subgroups in which the left kidney was innervated or denervated, respectively; in Group II, renal perfusion pressure (RPP) was maintained during the 2nd hypoxic period by occl97uding the distal aorta. 2. In both subgroups of Group I, both hypoxic periods produced hyperventilation, arterial PO2 falling to approximately 50 mmHg. Concomitantly, mean arterial pressure (MABP) fell by similar extents (approximately 23%, from a baseline level of 140 mmHg during the 2nd hypoxic period). In the innervated subgroup, renal vascular conductance (RVC) increased, but glomerular filtration rate (GFR) fell (by 48 and 6%, respectively, during the 2nd hypoxic period), while urine flow, absolute sodium excretion (UNaV) and fractional sodium excretion (FENa) fell (by 52, 63 and 61%, respectively). Baseline urine flow, UNaV and FENa were higher in the denervated subgroup, but hypoxia produced similar percentage changes from baseline in all variables. 3. In Group II, both subgroups showed similar changes during the 1st hypoxic period as the corresponding subgroups of Group I. However, during the 2nd hypoxic period when the fall in MABP was reduced to approximately 7%, the increase in RVC persisted only in the denervated subgroup; there was no significant change in GFR, urine flow, UNaV or FENa in either subgroup. 4. These results indicate that, in the rat, moderate hypoxia produces antidiuresis and antinatriuresis that are not dependent on the renal nerves, but are dependent on the hypoxia-induced fall in MABP. The fall in renal perfusion pressure (RPP) may directly determine renal function, but reflex influences upon the kidney initiated by, for example, arterial baroreceptor unloading, may play a role. The fall in GFR and increase in RVC, which persisted after denervation or when renal perfusion was controlled, implies a local dilatatory influence acting preferentially on the efferent arterioles.