The anxiolytic and memory-impairing effects of compounds that act at strychnine-insensitive (SI) glycine receptors were examined and compared with those of a competitive N-methyl-D-aspartate antagonist, 2-amino-7-phosphonoheptanoic acid (AP7); a use-dependent channel blocker, dizocilpine; and a benzodiazepine agonist, diazepam (DZP). Mice were trained to avoid a dark compartment and their latencies to step through were measured either within 1 hr after training in the presence of the drug (to assess the anxiolytic effects) or 24 hr after pre- or post-training treatment (to assess the effects on learning and memory). Post-training administration of the glycinergic compounds 1-aminocyclopropanecarboxylic acid, 7-chlorokynurenic acid and D-cycloserine reduced step-through latencies when testing was performed 30 min after drug treatment and within 1 h after training. Latencies were unaltered by these glycinergic compounds when testing was performed 24 hr later. Similar results were obtained with AP7 and DZP. In contrast, an amnesic dose of pentylenetetrazole reduced latencies both within 1 and 24 hr after training. Pretreatment with glycine abolished the reduction in latencies observed with SI glycine receptor ligands 1 hr after training but did not antagonize the reduction produced by AP7. Pretraining administration of SI glycine receptor ligands did not alter step-through latencies measured 24 hr later. In contrast, under these same conditions, AP7, dizocilpine and DZP produced a significant reduction in latencies. These results demonstrate that compounds that act at SI glycine receptors do not impair learning and memory at doses that are anxiolytic in a single-trial punishment paradigm.