High-strength bioactive glass-ceramic A-W was soaked in various acellular aqueous solutions different in ion concentrations and pH. After soaking for 7 and 30 days, surface structural changes of the glass-ceramic were investigated by means of Fourier transform infrared reflection spectroscopy, thin-film x-ray diffraction, and scanning electronmicroscopic observations, in comparison with in vivo surface structural changes. So-called Tris buffer solution, pure water buffered with trishydroxymethyl-aminomethane, which had been used by various workers as a "simulated body fluid," did not reproduce the in vivo surface structural changes, i.e., apatite formation on the surface. A solution, ion concentrations and pH of which are almost equal to those of the human blood plasma--i.e., Na+ 142.0, K+ 5.0, Mg2+ 1.5, Ca2+ 2.5, Cl- 148.8, HCO3- 4.2 and PO4(2-) 1.0 mM and buffered at pH 7.25 with the trishydroxymethyl-aminomethane--most precisely reproduced in vivo surface structure change. This shows that careful selection of simulated body fluid is required for in vitro experiments. The results also support the concept that the apatite phase on the surface of glass-ceramic A-W is formed by a chemical reaction of the glass-ceramic with the Ca2+, HPO4(2-), and OH- ions in the body fluid.