This study examines the relationship between mortality rates and hospital patient volume for major orthopaedic surgery. All Medicare patients from fiscal years 1993 and 1994 in diagnosis-related groups (DRGs) 209, 210, and 214 were included, covering hip and knee arthroplasty, other hip and femur procedures, and spine procedures. Within DRG 209, five procedures were studied in more detail: total hip arthroplasty, partial hip arthroplasty, revision total hip arthroplasty, total knee arthroplasty, and revision total knee arthroplasty. Higher-volume hospitals had lower mortality rates, both in-house and in-house plus 30-day, for each of the DRGs studied and for each of the individual procedures within DRG 209. Age and sex were examined as potential causes of the differences in mortality rates, but no attempt was made to adjust for comorbidities or orthopaedic degree of difficulty. Results for 1995 are included in an appendix.