The interface between bone and a bioactive glass cement--a mixture of bioactive glass powder and ammonium phosphate solution, previously reported on by the authors--was evaluated quantitatively and histologically. The materials tested were (1) the original bioactive glass cement (BCI cement); (2) an improved type of bioactive glass cement (BCII cement); (3) polymethylmethacrylate (PMMA) bone cement; and (4) a bioactive, apatite-wollastonite-containing, glass ceramic (A-WGC). Hardened cylindrical specimens of each cement were inserted loosely into canine femora and the interfacial shear strengths were measured using a push-out test. The interfacial strength values of the bioactive glass cements increased with prolonged implantation time. At each postimplantation time studied (8, 12, and 24 weeks), the interfacial strength value of BCI cement did not differ significantly from that of A-WGC. BCII cement interfacial strength was greater than that of BCI cement, whereas the interfacial strength of PMMA bone cement remained at a very low level throughout the study. Histological examinations revealed that direct bonding of both bioactive glass cements to bone had occurred without pathologic degradation. After 24 weeks, the defects between the bone and the bioactive glass cements had been filled with mature lamellar bone. Because the bioactive glass cement system developed by the authors, especially BCII cement, shows excellent osteoconductivity and bonds to bone tightly, we consider it to be a promising material for fixing prostheses into bone.