Despite progress in techniques, early or late deep infection develops in 1 to 5% of patients with prosthetic joint replacements. We report a retrospective review of 64 original frozen sections (FSs) compared with permanent sections, the preoperative clinical and intraoperative findings, and the subsequent culture results. A histologic section, frozen or permanent, was considered positive for infection if there were more than five polymorphonuclear leukocytes (PMNs) per high power field, excluding surface fibrin and inflammatory exudate, in at least five separate microscopic fields. The sensitivity of an FS as a diagnostic test to detect prosthetic infection when present was 43%. The specificity of an FS to correctly identify the absence of infection was 97%. These data support the conclusion that the FS is reasonably specific but not a sensitive diagnostic modality. If the preoperative evaluation is completely negative and the FS does not reveal PMNs, the reimplantation can be performed with minimal concern regarding the possibility of infection. If the preoperative and intraoperative evaluations are completely negative but PMNs are observed in the FS, then the surgeon should consider proceeding with caution; the decision to proceed with reimplantation might be delayed, depending on the surgeon's clinical assessment.