We examined the influence of the proportion of glass-ceramic powder in a bioactive bone cement of our formula on the bone-bonding ability of cement. Changes in cement bonding with time also were examined. The bioactive bone cement consisted of MgO-CaO-SiO2-P2O5-CaF2 glass-ceramic powder (AW-GC powder) and bisphenol-alpha-glycidyl methacrylate (Bis-GMA)-based resin. AW-GC powder was added to the cement as 0%, 30%, 50%, 70%, and 80% w/w. Rectangular plates (2 x 10 x 15 mm) of each cement with polished surfaces were implanted into the proximal metaphysis of the tibiae of male rabbits, and the failure load was measured by detaching tests 10 and 25 weeks after implantation. The failure loads of each cement were 0% = 0.03, 30% = 1.52, 50% = 2.67, 70% = 3.56, and 80% = 5.59 kg at 10 weeks, and 0% = 0.05, 30% = 1.68, 50% = 2.77, 70% = 3.80, and 80% = 6.37 kg at 25 weeks. Observation of the cement-bone interface revealed that all bioactive bone cements (30%-80%) formed direct contact with bone whereas intervening fibrous tissue was observed in all specimens of the 0% group. By scanning electron microscopy, all bioactive bone cements (30%-80% groups) showed direct contact with bone at the cement-bone interface. In the 0% group, direct contact with bone at the cement-bone interface was not observed. By electron-probe microanalysis, a Ca-P-rich layer was not detected at the cement-bone interfaces of the 30%-70% bioactive bone cements, but in some samples of the 80% cement specimens a thin Ca-P-rich layer (3 microns thick) was observed at the interface at 10 and 25 weeks after implantation. These results show that all of the bioactive bone cements tested had the ability to bond to bone and to function as bioactive composites of ceramics and polymers.