Former evaluations of the role of proteoglycans in mineralization have neglected to address the possibility that the metabolism of proteoglycans may be of significance in this regard. This problem was studied by using radiolabelling in vivo of rat calvaria with [35S]sulphate for 2-72 h and a sequential extraction procedure to yield two pools of newly synthesized proteoglycans: one obtained from non-mineralized tissue by extraction with guanidinium chloride (GdmCl) and another obtained only after demineralization with EDTA. Total radioactivity in calvaria was maximal after 12 h of incorporation, but by 36 h had declined to a level that was about 55-65% of maximum. Radioactivity in the GdmCl extract declined steadily after 12 h, whereas that in the EDTA extract remained constant until 36 h, when it began to increase. Each extract contained a minor proteoglycan that eluted at the void volume (V0) of a Sepharose CL-6B column. Unlike in the EDTA extract, this proteoglycan gradually disappeared from the GdmCl extract. Each extract also contained a major, smaller proteoglycan, with a K(av.) of 0.24 and 0.36 in the GdmCl and EDTA extracts respectively. Papain digestion of each extract yielded glycosaminoglycan chains with K(av.) values of 0.32 and 0.50 on CL-6B in the GdmCl and EDTA extracts respectively. Digestion of each extract with chondroitinase ABC and chondroitinase AC showed that the glycosaminoglycans were of similar disaccharide composition, with about 85% being 4-sulphated and the remainder 6-sulphated and/or iduronic acid-containing. These data suggest that about 45% of the newly synthesized proteoglycans are removed from the tissue during the course of mineralization.