The anticonvulsant properties of marijuana have been known for centuries. The recently characterized endogenous cannabinoid system thus represents a promising target for novel anticonvulsant agents; however, administration of exogenous cannabinoids has shown mixed results in both human epilepsy and animal models. The ability of cannabinoids to attenuate release of both excitatory and inhibitory neurotransmitters may explain the variable effects of cannabinoids in different models of epilepsy, but this has not been well explored. Using acute mouse brain slices, we monitored field potentials in the CA1 region of the hippocampus to characterize systematically the effects of the cannabinoid agonist WIN55212-2 (WIN) on evoked basal and epileptiform activity. WIN, acting presynaptically, significantly reduced the amplitude and slope of basal field excitatory postsynaptic potentials as well as stimulus-evoked epileptiform responses induced by omission of magnesium from the extracellular solution. In contrast, the combination of omission of magnesium plus elevation of potassium induced an epileptiform response that was refractory to attenuation by WIN. The effect of WIN in this model was partially restored by blocking γ-aminobutyric acid type B (GABA(B) ), but not GABA(A) , receptors. Subtle differences in models of epileptiform activity can profoundly alter the efficacy of cannabinoids. Endogenous GABA(B) receptor activation played a role in the decreased cannabinoid sensitivity observed for epileptiform activity induced by omission of magnesium plus elevation of potassium. These results suggest that interplay between presynaptic G protein-coupled receptors with overlapping downstream targets may underlie the variable efficacy of cannabinoids in different models of epilepsy.