Homocysteine thiolactone (HTL) elicits seizures in mice at a dose of 850 mg/kg (95-100% of animals) with an average latency time of 19.5 min. These seizures are reversed by both 5' N-ethylcarboximide adenosine (NECA) and flunitrazepam, with respective ED50 doses of 0.025 and 0.20 mg/kg. NECA was approximately four-fold more potent as an inhibitor of HTL-induced seizures than of seizures induced by pentylenetetrazol (PTZ, 75 mg/kg). Flunitrazepam was equipotent in both seizure paradigms. The purine precursor 5-amino-4-imidazole carboxamide riboside, (AICAr), although virtually ineffective against PTZ-induced seizures at doses greater than 1 g/kg, was able to inhibit HTL-induced seizures with an ED50 of approximately 350 mg/kg. The anticonvulsant effect of AICAr was dose and time dependent. The anticonvulsant potency of AICAr was increased by simultaneous administration of the adenosine uptake blocker Mioflazine, whereas the central nervous system (CNS)-impermeable adenosine uptake blocker dipyridamole had no effect. The ability of AICAr to permeate the blood-brain barrier (BBB) is limited (less than 1%) and may explain its low potency as an anticonvulsant. AICAr also has very low potency at brain adenosine A1 and A2 receptors as well as adenosine uptake sites (IC50 greater than 10(-3) M), suggesting that its anticonvulsant properties are not mediated by direct action at these sites. The results indicate that AICAr does have frank anticonvulsant effects and further suggest that HTL-induced seizures may represent a useful paradigm for evaluation of adenosinergic agents. AICAr or more potent derivatives thereof may represent a new class of anticonvulsants with the ability to target seizure foci selectively.