A hydraulic calcium phosphate cement with beta-tricalcium phosphate (TCP) granules embedded in a matrix of dicalcium phosphate dihydrate (DCPD) was implanted in experimentally created defects in sheep. One type of defect consisted of a drill hole in the medial femoral condyle. The other, partial metaphyseal defect was located in the proximal aspect of the tibia plateau and was stabilized using a 3.5 mm T-plate. The bone samples of 2 animals each per group were harvested after 2, 4, 6 and 8 weeks. Samples were evaluated for cement resorption and signs of immediate reaction, such as inflammation, caused by the cement setting in situ. Differences regarding these aspects were assessed for both types of defects using macroscopical, radiological, histological and histomorphometrical evaluations. In both defects the brushite matrix was resorbed faster than the beta-TCP granules. The resorption front was followed directly by a front of new bone formation, in which residual beta-TCP granules were embedded. Cement resorption occurred through (i) extracellular liquid dissolution with cement disintegration and particle formation, and (ii) phagocytosis of the cement particles through macrophages. Signs of inflammation or immunologic response leading to delayed new bone formation were not noticed at any time. Cement degradation and new bone formation occurred slightly faster in the femur defects.