objective. To compare the characteristics and risk factors for surgical site infections (SSIs) after total hip arthroplasty (THA) and total knee arthroplasty (TKA) in a nationwide survey, using shared case detection and recording systems. Design. Retrospective cohort study. Setting. Twenty-six hospitals participating in the Korean Nosocomial Infections Surveillance System (KONIS). Patients. From 2006 to 2009, all patients undergoing THA and TKA in KONIS were enrolled. Results. SSI occurred in 161 (2.35%) of 6,848 cases (3,422 THAs and 3,426 TKAs). Pooled mean SSI rates were 1.69% and 2.82% for THA and TKA, respectively. Of the cases we examined, 42 (26%) were superficial-incisional SSIs and 119 (74%) were "severe" SSIs; of the latter, 24 (15%) were deep-incisional SSIs and 95 (59%) were organ/space SSIs. In multivariate analysis, a duration of preoperative hospital stay of greater than 3 days was a risk factor for total SSI after both THA and TKA. Diabetes mellitus, revision surgery, prolonged duration of surgery (above the 75th percentile), and the need for surgery due to trauma were independent risk factors for total and severe SSI after THA, while male sex and an operating room without artificial ventilation were independent risk factors for total and severe SSI after TKA. A large volume of surgeries (more than 10 procedures per month) protected against total and severe SSI, but only in patients who underwent TKA. Conclusions. Risk factors for SSI after arthroplasty differ according to the site of the arthroplasty. Therefore, clinicians should take into account the site of arthroplasty in the analysis of SSI and the development of strategies for reducing SSI.