BACKGROUND: The relationship between volume and outcome of total knee arthroplasties has never been evaluated in a nationally representative sample, to our knowledge. We hypothesized that surgeons and hospitals with higher patient volumes would have better outcomes, as defined by lower mortality rates, shorter hospital stays, and lower postoperative complication rates. METHODS: The 1997 Healthcare Cost and Utilization Project (HCUP) Nationwide Inpatient Sample, Release 6, provided discharge abstracts of patients undergoing total knee arthroplasty from a national stratified probability sample. Logistic and multiple regression models were used to estimate the adjusted association of surgeon or hospital volume with rates of in-hospital mortality, pulmonary thromboembolism, deep venous thrombosis in the lower extremity, and postoperative wound infection as well as length of hospital stay. Estimates were calculated for a target population of 277,550 patients. Models were adjusted for comorbidity, age, gender, race, household income, and procedure (primary or revision arthroplasty). RESULTS: The patients were mostly white (70.2%) and female (62.7%), with a mean age of 68.9 years. The overall in-hospital mortality rate for the target population was 0.2%, and the average length of stay was 4.6 days for the primary total knee arthroplasties and 4.9 days for the revision procedures. Surgeon volumes of at least fifteen procedures per year and hospital volumes of at least eighty-five per year were significantly and linearly associated with lower mortality rates (odds ratio = 0.56 [0.24 to 1.31] for surgeon volume of > or = 60). No other association demonstrated a significant and directionally consistent linear trend for improved outcomes. CONCLUSION: Patients treated by providers with lower caseload volumes had higher rates of mortality following total knee arthroplasty in 1997. Proposing volume standards could decrease patient mortality following this procedure.