The reliability of combined indium-111 leukocyte/technetium-99m sulfur colloid scans, with and without the addition of blood pooling and blood flow studies, in the diagnosis of infected total joint arthroplasty was investigated. Both scans were performed on 58 patients before reoperation of total hip or knee arthroplasty in the period 1996-1999. Results for imaging alone included 100% specificity, 46% sensitivity, 100% positive predictive value, 84% negative predictive value, and 88% accuracy. Inclusion of blood pooling and flow phase data improved results to 66% sensitivity, 89% negative predictive value, and 90% accuracy, with reductions in specificity (98%) and positive predictive value (91%). Routine use of these radionuclide scans is not supported by these data.