The ability to identify intraoperatively patients with an infected prosthesis at the time of a revision procedure assists the surgeon in selecting appropriate management. The results of 413 intraoperative Gram stains were compared with the results of operative cultures, permanent histology, and the surgeon's intraoperative assessment to determine the ability of Gram stains to identify periprosthetic infection. Gram staining correctly identified the presence of infection in 10 of the 68 cases that met study criteria for infection (sensitivity of 14.7%). Four false-positive Gram stains were encountered. Intraoperative Gram stains do not have adequate sensitivity to be helpful in identifying periprosthetic infection and should not be performed on a routine basis. They may be helpful, however, in cases in which gross purulence is encountered to assist in the selection of initial antibiotic therapy. The use of intraoperative Gram staining alone is inadequate for ruling out infection at the time of revision total joint arthroplasty.