Background: The accurate preoperative diagnosis of infection is an essential component of decision-making prior to revision total knee arthroplasty. When preoperative modalities used to detect infection reveal equivocal findings, the surgeon may rely on intraoperative testing. While intraoperative Gram stains are routinely performed during revision total knee arthroplasty, their value remains unclear. Methods: We retrospectively reviewed the records on 945 revision total knee arthroplasties performed at three university institutions to which patients were referred for total joint arthroplasty; the results of an intraoperative Gram stain were available for review in 921 cases (97.5%). Of these knees, 247 were classified as infected on the basis of (1) the presence of the same organism in two cultures; (2) growth, on solid media, of an organism as well as other objective evidence of infection; (3) histologic evidence of acute inflammation; (4) gross purulence; and/or (5) an actively draining sinus. We reviewed the results of preoperative laboratory studies, which included measurements of the erythrocyte sedimentation rate, C-reactive protein values, and white blood-cell count in 90%, 76%, and 98% of cases, respectively. Preoperative aspiration to obtain a specimen for culture and a cell count was performed routinely at one center and selectively at the other two centers, and the results were available for review in 439 (48%) of the 921 cases. Results: Intraoperative Gram staining was found to have a sensitivity of 27% and a specificity of 99.9%. The positive and negative predictive values were 98.5% and 79%, respectively. The test accuracy was 80%. Patients with a truepositive Gram stain had a significantly higher preoperative white blood-cell count, C-reactive protein level, and nucleated cell count in the aspirate when compared with patients with a false-negative Gram stain (p < 0.001). In no case did the results of the intraoperative Gram stain alter treatment. Conclusions: The intraoperative Gram stain was found to have poor sensitivity and a poor negative predictive value, and its results did not alter the treatment of any patient undergoing revision total knee arthroplasty because of a suspected infection. These data do not support the routine use of intraoperative Gram staining in revision total knee arthroplasty; instead, they suggest a much more limited role for this test. Level of Evidence: Diagnostic Level I. See Instructions to Authors for a complete description of levels of evidence. Copyright © 2009 by The Journal of Bone and Joint Surgery, Incorporated.