We reviewed 194 revision arthroplasties of the hip and knee performed over a ten-year period. The results of intraoperative Gram staining were available in 169 (87%). Thirty-two were found to be infected (11 hips and 21 knees) and 137 had no evidence of infection.

Intraoperative Gram staining was negative in all 169 cases. The method therefore had a sensitivity of 0% for detecting infection. We conclude that the absence of organisms on intraoperative Gram staining during revision arthroplasty does not confirm the absence of infection.