Postoperative fever is often misinterpreted as a sign of infection, especially when occurring after the third postoperative day. We assessed the epidemiology of postoperative fever in adult orthopaedic patients and its association with infection. Among 1,073 patients participating in a prospective observational study, 198 (19%) had a postoperative fever (>38°C). Thirteen patients (1.2%) had a surgical site infection and 78 patients (7.3%) had remote bacterial infections during their hospital stay. Including asymptomatic bacteriuria, 174 patients were given antibiotic therapy for a median duration of six days. In multivariate analysis, no clinical parameter was associated with fever, including haematoma (odds ratio 0.9, 95%CI 0.6-1.3), infection (1.6, 0.7-3.7), or antibiotic use (1.6, 0.9-3.0). The maximum temperature on each of the first seven postoperative days did not differ between infected and uninfected patients (Wilcoxon rank-sum tests; p > 0.10). We conclude that fever, even up to the seventh postoperative day, is not substantially helpful to distinguish infection from general inflammation in clean orthopaedic surgery.