The pharmacologic specificity of tolerance to caffeine-induced stimulation of locomotor activity was studied in adult male rats that were given access to either caffeine solution (0.5 or 1.0 mg/ml) or plain water for 10 min every 6 h on a chronic daily basis; daily caffeine intake averaged 41 and 62 mg/kg, respectively. Dose-effect curves were determined for behavioral stimulant and depressant drugs in control and caffeine-treated groups. Drugs were injected IP and locomotor activity was measured for 30 min beginning 35 min later. Rats tolerant to stimulation of locomotor activity by caffeine were also tolerant to theophylline and 7-(2-chloroethyl)theophylline, but not to any of six nonxanthine stimulants, including cocaine, methylphenidate, and d-amphetamine. The adenosine analogs, R(-)-N6-2-(phenylisopropyl)adenosine(R(-)-PIA) and 5'-(N-ethyl)carboxamidoadenosine (NECA), decreased locomotor activity of control and caffeine-treated (0.5 mg/ml) rats; dose-effect curves in rats consuming caffeine chronically were displaced to the right of the control curves by 10-fold for R(-)-PIA and 100-fold for NECA. Dose-effect curves for the nonadenosine behavioral depressants chlorpromazine and diazepam were unchanged by chronic treatment with caffeine, but the curve for pentobarbital, which is thought to inhibit adenosine receptor binding, was shifted to the right by a factor of 3. Rats withdrawn from chronic caffeine for 24 h were still completely tolerant to caffeine-induced stimulation of locomotor activity. Dose-effect curves for R(-)-PIA and d-amphetamine in rats withdrawn from chronic caffeine for 24 h were not different from curves in control animals.(ABSTRACT TRUNCATED AT 250 WORDS)