Pivampicillin was administered as an oral paste to five healthy adult horses, and an oral paste with ampicillin trihydrate was administered to three horses. Pivampicillin was administered to both starved and fed horses, ampicillin trihydrate was administered to fed horses only. The dose of pivampicillin was 19.9 mg/kg, and the dose of ampicillin trihydrate was 17 mg/kg. Both doses are equivalent on a molecular basis to 15 mg/kg ampicillin. Ampicillin concentrations in plasma were determined up to 24 hours after administration. After administration of pivampicillin to starved and fed horses the mean areas under the plasma concentration-time curve (AUCs) were 23.0 and 19.3 micrograms.h.ml-1, respectively. After administration of ampicillin trihydrate to fed horses the mean AUC was 0.7 microgram.h.ml-1. The peak plasma concentrations were 4.8, 6.7, and 0.1 microgram/ml, after administration of pivampicillin to starved and fed horses and of ampicillin trihydrate to fed horses, respectively. There was no statistically significant difference in peak plasma concentration or AUC between pivampicillin administered to starved or fed horses. It is concluded that pivampicillin administered as an oral paste at a dose of 19.9 mg/kg gives satisfactory plasma concentrations in both starved and fed horses, whereas ampicillin trihydrate produces negligible plasma concentrations. Pivampicillin binds to feedstuffs at the pH found in the horse's stomach and small intestine. After incubation for 6 h at pH 6, approximately 15% remains in solution, and after incubation for 3 h at pH 1.9, approximately 40% remains in solution. Ampicillin, which binds to feedstuffs to a lesser extent, has a lower bioavailability than pivampicillin. Therefore, binding to feedstuffs does not seem to be a critical factor in the absorption of aminopenicillins.