A hydraulic calcium phosphate cement made of beta-tricalcium phosphate [beta-Ca3(PO4)2], monocalcium phosphate monohydrate [Ca(H2PO4)2-H2O], and water was used as a delivery system for the antibiotic gentamicin sulfate (GS). GS, added as powder or as aqueous solution, was very beneficial to the physicochemical properties of the cement. The setting time increased from 2 to 4.5 min with 3% (w/w) GS and then slowly decreased to 3.75 min with 16% (w/w) GS. The tensile strength increased from 0.4 to 1.6 MPa with 16% (w/w) GS. These effects were attributed to the presence of sulfate ions in GS. The release of GS from the cement was measured in a pH 7.4 phosphate-buffered saline solution at 37 degrees C by USP paddle method. Factors such as cement porosity, GS content and presence of sulfate ions or polymeric additives were investigated. The amount of GS released was roughly proportional to the square root of time up to approximately 50% release. Afterwards, the release rate markedly slowed down to zero. In all but two cement formulations, the total dose of GS was released within 7 days, indicating that no irreversible binding occurred between the cement paste and the antibiotic. When small amounts of hydroxypropylcellulose or poly(acrylic acid) were added to the cement, the maximum fraction released was a few percent lower than the total GS dose, suggesting some binding between the polymer and GS. The GS release rate was strongly influenced by the presence of sulfate ions in the cement paste and by the cement porosity. The higher the sulfate ion content of the cement paste, the lowe the GS release rate. This influence was attributed to the finer cement micro-structure induced by the presence of sulfate ions. Furthermore, when the initial cement porosity was increased from 38 to 69%, the release rate almost tripled (0.16 to 0.45 h-1/2). Finally, the biological activity of GS in the cement was maintained, as measured by assaying the release medium.