Mineralization of implanted bioprostheses poses a major clinical problem. Crosslinking of collagenous matrices, a process used to render tissues relatively inert and nonbiodegradable, seems to encourage calcification. Residual, noncovalently bound glutaraldehyde, as well as glutaraldehyde crosslinks which can be degraded with time, seem to play a role in this connection. Our findings demonstrate the need to carefully remove noncovalently or labile-associated glutaraldehyde by thorough rinsing or neutralization before implantation. Components of a valve prosthesis such as cusps and aortic wall, which are known to vary in their proportions of collagen, elastin, and noncollagenous proteins and to calcify to different extents, can both be prevented from calcifying if treated with a biphosphonate before implantation. Calcification can also be reduced by selective enzymatic removal of noncollagenous materials. In addition to the age of rats, animals usually used to evaluate calcification, the strain of animal can markedly affect the response. The Fischer-344 rat, a highly inbred animal, will not calcify exhaustively rinsed implants. Our findings suggest that multifactorial approaches may have to be combined to generate the most ideal bioprostheses. These should include careful removal of noncovalently bound glutaraldehyde, neutralization of the nonbifunctionally reacted residues, removal of lipids and noncollagenous proteins (and possibly the more antigenic nonhelical collagen telopeptides), as well as inclusion of agents such as biphosphohates, which by interfering with crystal growth prevent the accumulation of mineral in the interstices of the tissue.