1. The question of whether active uptake limits the duration of action of the inhibitory transmitter glycine has been addressed in vivo at inhibitory synapses on the goldfish Mauthner (M) cell. The kinetics of inhibitory postsynaptic potentials and inhibitory postsynaptic currents (IPSCs) evoked antidromically and by eighth-nerve stimulation were recorded in control and in conditions expected to block glycine uptake or slow its diffusion. Theoretical considerations, based on simulated quantal currents, predicted that if diffusion was slow, rapid uptake of glycine would be required and its block would prolong the synaptic responses. 2. Temperature coefficient values for IPSC decay time constants (tau S) are in the range of 2.0 for temperatures between 15 and 23 degrees C, suggesting that diffusion is not the rate-limiting step. 3. Li+, Ch+, or N-methyl-D-glucamine were substituted for 80% of the Na+ in the extracellular fluid to analyze the effects of blocking the Na(+)-dependent glycine uptake. These procedures enhanced the maximum inhibitory shunt produced by glycine iontophoresis, leading to the suggestion that uptake may buffer the concentration of the transmitter in the cleft. In contrast, the Na+ substitutes had no effect on the tau of the recurrent collateral IPSC, which involves synchronous activation of a pool of interneurons and has a monoexponential decay (tau approximately 10-11 ms). 4. The decay phase of the disynaptic IPSCs produced by stimulating the contralateral eighth nerve has fast and slow components, with a prolonged tail lasting up to 100 ms, particularly in the case of repetitive nerve stimulation. The tail is inhibitory, as revealed by its shunt of the antidromic action potential, and it is at least partially Cl- dependent. However, it can be accelerated by superfusion with the glutamate receptor antagonists 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) and DL-2-amino-5-phosphonopentanoic acid (APV). In the presence of these blockers, the IPSC decay remains biexponential (tau fast = 5.2 and 5.9 ms, tau slow = 94 and 130 ms for single and burst stimuli, respectively). Blocking uptake in this condition did not modify tau fast or tau slow. 5. We conclude that an active uptake mechanism does not shape glycinergic IPSCs, including the longer-lasting components that might include a contribution due to persistence of the transmitter. Rather, diffusion alone is sufficient to remove glycine at a rate faster than channel unbinding. 6. To test whether glycine might diffuse to adjacent excitatory synapses and enhance activation of N-methyl-D-aspartate receptors, CNQX and APV were applied locally, by pressure, to the M cell soma, but they had no effect on the prolonged decay of eighth-nerve-evoked responses. Thus the effects of the antagonists when added to the superfusate are exerted at the network level.