BACKGROUND: Surgical treatment of peripheral vascular disease (PVD) in dialysis patients is controversial. METHODS: We examined the post-operative morbidity and mortality of surgical revascularization or amputation for PVD in a retrospective analysis of United States Renal Data System. Propensity scores for undergoing amputation were derived from a multivariable logistic regression model of amputation. RESULTS: Of the Medicare patients initiated on dialysis from Jan 1, 1995 to Dec 31, 1999, patients underwent surgical revascularization (n = 1,896) or amputation (n = 2,046) in the first 6 months following initiation of dialysis were studied. In the logistic regression model, compared to claudication, presence of gangrene had a strong association with amputation [odds ratio (OR) 19.0, 95% CI (confidence interval) 13.86-25.95]. The odds of dying within 30 days and within 1 year were higher (30 day OR: 1.85, 95% CI: 1.45-2.36; 1 yr OR: 1.46, 95% CI: 1.25-1.71) in the amputation group in logistic regression model adjusted for propensity scores and other baseline factors. Amputation was associated with increased odds of death in patients with low likelihood of amputation (< 33rd percentile of propensity score) and moderate likelihood of amputation (33rd to 66th percentile) but not in high likelihood group (> 66th percentile). The number of hospital days in the amputation and revascularization groups was not different. CONCLUSION: Amputation might be associated with higher mortality in dialysis patients. Where feasible, revascularization might be preferable over amputation in dialysis patients.