BACKGROUND: The optimal management of low-gradient "severe" aortic stenosis (mean gradient <40 mm Hg, indexed aortic valve area ≤0.6 cm2/m2) with preserved left ventricular ejection fraction remains controversial because gradients may be similar after aortic valve replacement (AVR). We compared outcomes of low-gradient severe aortic stenosis with AVR or medical therapy. METHODS AND RESULTS: Comprehensive echocardiographic measurements including hemodynamic calculations were completed in 260 prospectively identified patients with symptomatic low-gradient severe aortic stenosis. Patients were followed up for mortality over 28±24 months. AVR was performed in 123 patients (47%). Compared with AVR patients, medically treated patients had a higher prevalence of diabetes mellitus (25% versus 41%, P=0.009), lower stroke volume index (36.4±8.4 versus 34.4±8.7 mL/m2, P=0.02), higher pulmonary artery pressure (38±11 versus 48±21 mm Hg, P=0.001), and higher creatinine level (1.1±0.4 versus 1.22±0.5 mg/dL, P=0.02). These and other clinically relevant variables were entered into a propensity model that reflected likelihood of referral to AVR. This score and other variables were entered into a Cox model to explore the independent effect of AVR on outcome. During follow-up, 105 patients died (40%): 32 (30%) in the AVR group and 73 (70%) in the medical treatment group. AVR (hazard ratio, 0.54; 95% confidence interval, 0.32-0.94; P<0.001) was independently associated with outcome and remained a strong predictor of survival after adjustment for propensity score. Medical therapy was associated with 2-fold greater all-cause mortality than AVR. The protective effect of AVR was similar in 125 patients with normal flow (stroke volume index >35 mL/m2; P=0.22). CONCLUSIONS: AVR is associated with better survival than medical therapy in patients with symptomatic low-gradient severe AS and preserved left ventricular ejection fraction.