IMPORTANCE Shared decision making helps patients and clinicians elect therapies aligned with patients' values and preferences. This is particularly important for invasive therapies with considerable trade-offs. OBJECTIVE To assess the effectiveness of a shared decision support intervention for patients considering destination therapy left ventricular assist device (DT LVAD) placement. DESIGN, SETTING, AND PARTICIPANTS From 2015 to 2017, a randomized, stepped-wedge trial was conducted in 6 US LVAD implanting centers including 248 patients being considered for DT LVAD. After randomly varying time in usual care, sites were transitioned to an intervention consisting of clinician education and use of DT LVAD pamphlet and video patient decision AIDS. Follow up occurred at 1 and 6 months. MAIN OUTCOMES AND MEASURES Decision quality as measured by knowledge and values-choice concordance. RESULTS In total, 135 patients were enrolled during control and 113 during intervention periods. At enrollment, 59 (23.8%) participants were in intensive care, 60 (24.1%) were older than 70 years, 39 (15.7%) were women, 45 (18.1%) were racial/ethnic minorities, and 62 (25.0%) were college graduates. Patient knowledge (mean test performance) during the decision-making period improved from 59.5%to 64.9% in the control group vs 59.1%to 70.0%in the intervention group (adjusted difference of difference, 5.5%; P = .03). Stated values at 1 month (scale 1 = "do everything I can to live longer..." to 10 = "live with whatever time I have left...") were a mean of 2.37 in control and 3.33 in intervention (P = .03). Patient-reported treatment choice at 1 month favored LVAD more in the control group (than in the intervention group (47 [59.5%] vs 95 [91.3%], P < 001). Correlation between stated values and patient-reported treatment choice at 1 month was stronger in the intervention group than in the control group (difference in Kendall's τ, 0.28; 95%CI, 0.05-0.45); however, there was no improved correlation between stated values and actual treatment received by 6 months for the intervention compared with the control group (difference in Kendall's τ, 0.01; 95%CI, -0.24 to 0.25). The adjusted rate of LVAD implantation by 6 months was higher for those in the control group (79.9%) than those in the intervention group (53.9%, P = .008), with significant variation by site. There were no differences in decision conflict, decision regret, or preferred control. CONCLUSIONS AND RELEVANCE A shared decision-making intervention for DT LVAD modestly improved patient decision quality as measured by patient knowledge and concordance between stated values and patient-reported treatment choice, but did not improve concordance between stated values and actual treatment received. The rate of implantation of LVADs was substantially lower in the intervention compared with the control group.