The release of gentamicin sulphate, sodium fusidate and diethanolamine fusidate from Palacos and CMW cements was studied using elution and serial plate transfer tests. Further tests were made to assay the drug remaining in the cement after antibacterial activity could no longer be detected by the above methods, to detect the sustained slow release of the residual drug, and to ascertain the mechanism of release. The results confirmed that the release of gentamicin sulphate could be detected for longer from Palacos cement than from CMW cement, but the opposite was true for sodium fusidate. Little difference was found in the case of diethanolamine fusidate. Comparison of elution and serial plate transfer tests, and of results of elution in buffers of different pH, demonstrated that the test method employed had a significant effect on the results, and the omission of details of methodology from some publications made comparison and evaluation of results difficult. Varying quantities of residual drug were found in cement from which antibacterial activity could no longer be demonstrated; further tests for sustained, slow release showed that the antibiotic did not remain fixed in the cement but was released at a rate too slow to be detected in the elution and serial plate transfer tests. It is concluded that antibiotics are released from the cement by a process of diffusion, but tests to determine the mechanism of diffusion were unhelpful. The theory of diffusion of drugs through solid matrices, and the clinical implications of the experimental findings, are discussed.