A total of 202 revision hip arthroplasties in 178 patients, over a 2-year period, were evaluated prospectively. Intraoperative Gram stains were obtained from periprosthetic tissues in all cases. Of these, a definitive diagnosis of infection, using defined criteria, was established in 35 cases. Of these 35 patients, 17 had received antibiotics before surgery. The intraoperative cultures were positive in 8 of the 17 patients who had received antibiotics and 17 of the 18 patients who had not received preoperative antibiotics. In 1 infected case, intraoperative cultures of periprosthetic tissues failed to reveal bacterial growth, despite the fact that the patient received no preoperative antibiotics. Overall, there were 5 true-positive Gram stain results, 172 true-negative results, 3 false-positive results, and 22 false-negative results. The sensitivity of the Gram stain was 19%, specificity was 98%, predictive value of a positive test was 63%, and predictive value of a negative test was 89%. These results suggest that the intraoperative Gram stain is not a sensitive tool for the diagnosis of infection and should not be used when attempting to diagnose infection intraoperatively.