Seventy-two joint arthroplasties undergoing total hip or total knee surgery were studied prospectively with plain radiographs, three-phase bone imaging (3PBI), erythrocyte sedimentation rate (ESR), aspiration of the joint for culture, and multiple intraoperative cultures at the time of revision. Intraoperative cultures and the operative appearance were used to form a diagnosis of definite infection (unequivocal microbiology and gross sepsis), possible infection (positive microbiology or gross sepsis), or no infection (neither positive microbiology nor gross sepsis). For the preoperative diagnosis of infection, as opposed to aseptic loosening, 3PBI alone had a sensitivity of 33% and a specificity of 86%. In conjunction with plain radiographs, minimal improvement in accuracy was seen. A preoperative ESR greater than 30 had low sensitivity (60%) and a specificity of (65%). However, the ESR was statistically significantly higher in the joints with definite infection as compared to those joints without infection. The preoperative joint aspiration had a sensitivity of 67% and a specificity of 96% and, therefore, appears to be the most useful single test in the workup of a painful total joint arthroplasty.