An aqueous solution and a lipid emulsion of bupivacaine were administered epidurally in doses of 1.8 mg/kg to six beagle dogs following a randomised two-phase crossover design. The aqueous solution was absorbed rapidly and the mean (sd) peak venous plasma concentration of bupivacaine, 1.4 (0.4) μg/ml, was detected after five minutes. After administration of the lipid emulsion, the peak plasma concentration of bupivacaine, 0.6 (0.2) μg/ml, was detected after 30 minutes. The mean (sd) t½β of the aqueous preparation was 149.1 (32.6) minutes, and of the lipid emulsion 119.2 (34.0) minutes. Both preparations had a similar bioavailability. The mean time to the onset of motor block after the administration of the aqueous solution, 2.3 (2.2) minutes, was significantly shorter (P=0.028) than after the administration of the lipid emulsion, 9.4 (1.9) minutes, and the duration of the motor block induced by the lipid emulsion, 217.6 (26.2) minutes, was significantly longer (P=0.043) than for the aqueous solution, 158 (48.8) minutes. During anaesthesia, the plasma concentrations of bupivacaine ranged between 1.3 and 0.2 μg/ml. Non-significant changes in systolic blood pressure and heart rate were observed which coincided with the peak plasma concentrations of bupivacaine.