The anxiogenic action of caffeine (10, 25 and 50 mg/kg, i.p.) was investigated in rats and compared with that of yohimbine (2 mg/kg, i.p.). The experimental methods used were the open-field, elevated plus-maze, social interaction and novelty-suppressed feeding latency tests. Caffeine produced a dose-related profile of behavioural changes, which were qualitatively similar to those induced by yohimbine and which indicate an anxiogenic activity in rodents. Thus, both the drugs reduced ambulation and rears, and increased immobility and defaecation in the open-field test. They decreased the number of entries and time spent on the open arms of the elevated-plus maze, reduced social interaction in paired rats and increased the feeding latency in an unfamiliar environment in 48-h food-deprived rats. Lorazepam, a well known benzodiazepine anxiolytic agent, attenuated the anxiogenic effects of caffeine and yohimbine. Subchronic administration of caffeine (50 mg/kg, i.p.) for 21 days, in different groups of animals, induced a significant degree of tolerance in the elevated plus-maze test, which was statistically significant after 14 and 21 days' treatment. Yohimbine, however, did not induce similar tolerance. When caffeine (50 mg/kg, i.p.) was withdrawn after 21 days' administration, to a separate group of rats, significant withdrawal anxiety was observed 48 h later as noted in the elevated plus-maze test. The investigations support clinical evidence of caffeine-induced anxiety, tolerance to anxiety on continued use, and withdrawal anxiety in chronic caffeine-containing beverage users.