A gentamicin carrier system composed of calcium phosphates, poly(DL-lactide) (PLA) and gentamicin was developed and characterized in vitro and in vivo for use in the prevention and treatment of bone infection. Four formulations were prepared according to an experimental design based on the Hadamard matrix. The technological variables included in the design were: gentamicin loading with respect to the implant weight, weight average molecular weight (M(w)) of the PLA as a compound of the matrix and the presence or absence of a PLA coating of 200 kDa. The variable to be optimized in vitro was the gentamicin release level during the first week. According to this goal, the selected formulation was F-D which was composed of 80% phosphates (25% hydroxyapatite, HAP and 75% tricalcium phosphate, TCP), 20% PLA (M(w), 30 kDa) and 3.5% gentamicin sulfate (GS) and was coated with PLA (M(w), 200 kDa). To elucidate the in vitro release mechanism of this implant, another implant lot (F-X) uncoated, but with identical matrix composition, was prepared. Results showed that the PLA coating delay the gentamicin release, indicating that part of the antibiotic released from the matrix diffuses through the polymer coating film. The selected formulation was tested in the femur of rabbits and showed a faster release rate in vivo than in vitro. This is due to a greater degree of PLA degradation, changes in the phosphate blend, and bone tissue invading the implant. Gentamicin concentration in the areas of the bone closest to the implant was higher than the minimum inhibitory concentration (MIC) against Staphylococcus aureus.