Ninety patients who had undergone 92 total knee replacements were reviewed to determine predictors of postoperative pyrexia and to evaluate the relative value of septic screening in this group of patients. Postoperative pyrexia was defined as an axillary temperature greater than 37 degrees C (98.4 degrees F) on any or all of the 5 days after surgery. All of the patients in this series developed a postoperative pyrexia after knee arthroplasty. None of 16 patients (17% of arthroplasties) with a temperature of 39 degrees C (102 degrees F) or greater had evidence of infection. None of the 4 patients with urinary tract infections developed a pyrexia exceeding 38 degrees C (101 degrees F). At a minimum of 2 years' follow-up, none of the 90 patients had developed an infected arthroplasty. Logistic regression analysis showed that the risk of becoming significantly febrile (temperature > 39 degrees C) doubled for every unit drop in hematocrit and increased fourfold for each unit of blood transfused after surgery. Perioperative blood loss and pyrexia were correlated, but the correlation was not statistically significant. The duration of procedure, tourniquet time, and patient age or sex were not associated with risk of pyrexia. In this series, there was no association between a pyrexia greater than 37 degrees C, septic screening, and the presence of an infective focus. Early postoperative pyrexia after arthroplasty is a normal physiological response, and a significant pyrexia can be predicted by a drop in hematocrit and/or after postoperative transfusion. Pyrexia in the early postoperative period following total knee arthroplasty warrants detailed laboratory and radiographic investigation only in the presence of positive physical findings.