OBJECTIVES: Evaluation of peripheral vascular disease in the primary care setting is routinely performed by contrast-enhanced magnetic resonance angiography (CE-MRA) and digital subtraction angiography (DSA). However, limited data are available on the relative costs and clinical outcomes following these diagnostic procedures. METHODS: We identified individuals who underwent an outpatient CE-MRA (3,444) or DSA (16,899) procedure of the lower extremities from 1998 to 2004 in the U.S. Veterans Affairs system. Using VA costing algorithms, Cost and log-cost of interventions (e.g., revascularization, stent, angioplasty), amputations or mortality rates within 30 days and one year of DSA or CE-MRA were compared for both groups adjusted for baseline characteristics using multivariate regression. Imaging modality selection bias was evaluated with propensity score, instrumental variables and Heckman methods using retransformation and Generalized Linear Models. RESULTS: Initial CE-MRA imaging was significantly more likely among patients with prior renal disease or bypass surgery (OR >2; p<0.001) and less likely among patients with prior amputation, PVD, claudication or other cardiovascular disease markers (OR <0.7; p<0.001). Even after adjusting for endogenous choice of initial imaging modality, 30 day treatment costs were US$3,500-$4,300 lower (p<.001) for imaging patients with initial CE-MRA. Eighty-two percent (65%) of DSA imaging patients had no additional procedures or events within 30 (90) days, and less than 3.2% (3.6%) of patients had any repeat image within 30 (90) days of their initial image. CONCLUSION: Relative to DSA, CE-MRA imaging was associated with substantial treatment episode savings, beyond the US$950 lower imaging cost per procedure. These savings estimates were robust to alternate model specifications and statistical corrections for endogenous imaging choice. Substituting CE-MRA for DSA among those not planning or requiring any follow-up procedures within 30 days could have reduced outpatient imaging costs by up to 55%, and reduced VA system costs by US$13.2 million.