The pharmacokinetics of caffeine (CAF) and its metabolites, dimethylxanthines, were examined in horses administered 2.5 mg/kg of CAF intravenously (i.v.), intramusculary (i.m.), or orally (p.o.). The plasma samples were extracted by Extrelut^[○!R] and the concentrations of CAF and metabolites were determined by high performance liquid chromatography (HPLC) with a short column. The pharmacokinetics of CAF after bolus i.v. injection were described by the assumption of a two-compartment model, and those of CAF after i.m. or p.o. administration were done by the assumption of a one-compartment model. The biologic half lives of CAF were 15.5, 18.6, and 16.4 h after administering i.v., i.m. and p.o., respectively. The extent of the bioavailability of the p.o. administration was determined as 1.04 times the dose. The differences in pharmacokinetic parameters were not statistically significant among administration routes. A straight correlation existed between the logarithms of body weights of different species of animals and those of their biologic half lives of CAF. Therefore, the biologic half life of CAF in an animal might be predictable as a function of its body weight.