The effect of procedure duration on joint arthroplasty survivorship in the USA is unknown. We examined the association between procedure duration with primary total hip arthroplasty and total knee arthroplasty survivorship at 8 years in the Medicare population using 1997 to 2004 Medicare claims data. Procedure duration was determined using anesthesia time as a proxy. Kaplan-Meier analysis and Cox regression were used to determine implant survivorship at 8 years and hazard ratios associated with procedure duration. Total knee arthroplasty implant survival was significantly associated with procedure duration (P = .001), in contrast to total hip arthroplasty (P = .127). Total knee arthroplasty procedures shorter than 90 minutes, between 150 and 180 minutes, and more than 240 minutes had significantly higher revision rates than those lasting 120 to 150 minutes. Total hip arthroplasty procedures lasting more than 240 minutes also had a significantly higher revision risk than those lasting 120 to 150 minutes. Our findings support the general belief that longer procedures are associated with the greater probability of complications.