The relationship between in vitro and in vivo indomethacin (IMC) release from a self-setting bioactive apatite cement and cement size were investigated. Differently sized apatite cements (total weight, 500 mg); either 64 of the small size (2 mm diameter x 2 mm thickness), sixteen of the medium size (4 mm x 2 mm) or one of the large size (15 mm x 2 mm) were obtained from cement bulk powder containing tetracalcium phosphate, dicalcium phosphate dihydrate and hydroxyapatite. In vitro IMC release from the 1, 2 and 5% drug-loaded apatite cement systems in simulated body fluid (SBF) (pH 7.25) at 37 degrees C increased with increasing concentrations of IMC and with decreasing geometrical size of the cement. The plots of in vitro IMC release per unit area against the square root of time increased with increasing IMC concentrations, but not with decreasing geometrical size of the cement. After subcutaneous (s.c.) implantation of differently sized 1% IMC-loaded cements in male Wistar rats, the plasma IMC concentration and the area under the curve increased with decreasing cement diameter. The in vivo IMC release profiles of the cement were deconvoluted from the plasma IMC profiles after s.c. administration of IMC solution. The plots of in vivo IMC release per unit area against the square root of time suggested that the initial release from all 1% drug-loaded cements was very rapid, slowed after one day, but continued for over two weeks. The relationship between the in vitro release in SBF and the in vivo release in rats of IMC-loaded cements was linear.