The pharmacokinetics of amoxicillin (AMX) were investigated in sheep following intravenous (i.v.) and intramuscular (i.m) injection, comparing two different drug formulations, a conventional and a long-acting AMX-trihydrate suspension. For the i.m. application two different injections sites, the neck area and the hind limb were used to identify possible differences in the kinetic parameters related to the site of injection. A three-compartment open model could best describe AMX disposition after i.v. administration. Data analysis after i.m. administration of the conventional suspension at both injection sites revealed the occurrence of a flip-flop phenomenon, clearly indicating that absorption of AMX is the rate-limiting step of its overall disposition. A moderate effect of the injection site was observed with a tendency for the neck area to be advantageous, mainly in terms of rate rather than extent of absorption. Injection of the long-acting formulation led to a focal depot formation, thus yielding lower but remarkably prolonged serum AMX levels reflected in the respective terminal half-lives. The concentration-time profile of AMX after administration of the long-acting formulation was less affected by the injection site, but the low serum levels justify its use only in cases in which a high susceptibility of the involved bacterial population is confirmed.