The in vivo activity of phosphate-activated glutaminase (PAG) was measured in the brain of hyperammonaemic rat by 15N n.m.r. Brain glutamine was 15N-enriched by intravenous infusion of 15NH4+ until the concentration of [5-15N]glutamine reached 6.1 mumol/g. Further glutamine synthesis was inhibited by intraperitoneal injection of methionine-DL-sulphoximine, an inhibitor of glutamine synthetase, and the infusate was changed to 14NH4+ during observation of decrease in brain [5-15N]glutamine due to PAG and other glutamine utilization pathways. Progressive decrease in brain [5-15N]glutamine, PAG-catalysed production of 15NH4+ and its subsequent assimilation into glutamate by glutamate dehydrogenase were monitored in vivo by 15N n.m.r. Brain [5-15N]glutamine (15N enrichment of 0.35-0.50) decreased at a rate of 1.2 mumol/h per g of brain. The in vivo PAG activity, determined from the observed rate and the quantity of 15NH4+ produced and subsequently assimilated into glutamate and aspartate, was 0.9-1.3 mumol/h per g. This activity is less than 1.1% of the reported activity in vitro measured in rat brain homogenate at a 10 mM concentration of the activator Pi. Inhibition by ammonia (brain level 1.4 mumol/g) alone does not account for the observed low activity in vivo. The result strongly suggests that, in intact brain, PAG activity is maintained at a low level by a suboptimal in situ concentration of Pi and the strong inhibitory effect of glutamate. The observed PAG activity in vivo is lower than the reported in vivo activity of glutamate decarboxylase which converts glutamate into gamma-aminobutyrate (GABA). The result suggests that PAG-catalysed hydrolysis of glutamine is not the sole provider of glutamate used for GABA synthesis.