1. Kynurenate (Kyn), 7-chlorokynurenate (7-Cl-Kyn), 3-amino-1-hydroxypyrrolid-2-one (HA-966) and D-cycloserine are known to bind to the glycine site that modulates the N-methyl-D-aspartate (NMDA) response of vertebrate central neurones. The effects of these compounds were investigated with patch-clamp and fast-perfusion techniques on mouse cortical neurones in primary culture in an effort to establish whether they act as antagonists, partial agonists and/or inverse agonists of glycine. A fast drug application method allowed the study of both steady-state and transient responses. 2. The analysis of steady-state responses indicates that the main effects of Kyn and 7-Cl-Kyn are those expected from competitive antagonists of glycine, with a dissociation constant of 15 microM for Kyn, and of 0.3 microM for 7-Cl-Kyn. Concentration jumps indicate that at all concentrations of glycine, and in particular in the absence of added glycine, the blockade by Kyn and 7-Cl-Kyn develops at a rate which is close to the rate of dissociation of glycine from its binding site and is independent of antagonist concentration. 3. The main effects of D-cycloserine and of HA-966 are those of partial agonists of high and low efficacy, respectively. In the absence of added glycine, D-cycloserine always produced a potentiation, while HA-966 produced either a potentiation or an inhibition. This can be explained by assuming the presence of a variable level of contaminating glycine. With both D-cycloserine and HA-966, concentration jumps produced biphasic relaxations in which the onset rate of the slow component was, here again, close to the rate of dissociation of glycine from its binding site. 4. These results can be interpreted by assuming that (1) Kyn and 7-Cl-Kyn are competitive antagonists of glycine, (2) HA-966 and D-cycloserine are partial agonists, (3) in the absence of added glycine some glycine is present in the extracellular solution and (4) the response in the total absence of glycine is very small or negligible.