Older adults with multiple comorbidities are often undernourished or at high risk for becoming so, especially after a recent hospitalization. Randomized controlled trials of effective, innovative interventions are needed to support evidence-based approaches for solving nutritional problems in this population. Self-management approaches where participants select their own behavioral goals can enhance success of interventions. The purpose of this study was to evaluate the feasibility and efficacy of a multilevel self-management intervention to improve nutritional status in a group of high-risk older adults. The Behavioral Nutrition Intervention for Community Elders (B-NICE) trial used a prospective randomized controlled design to determine whether the intervention, compared to standard care, maintained or increased caloric intake (depending on baseline body mass index) and, consequently, stabilized or increased body weight. Participants were 34 Medicare-eligible, age 65 years old or older, homebound adults who were consuming insufficient calories and/or had a history of weight loss ≥2.5% over 6 months. The intervention took place within participants' homes. Outcome measures, including energy intake (based on collection of three 24-hour dietary recalls) and body weights were assessed at baseline and at 60 days post randomization. The primary analyses included analyses of covariance and Pearson's χ2. We hypothesized that the intervention would result in increased caloric intake and weight gain in underweight older adults and increased or stabilized caloric intake and weight for everyone else. The intervention was feasible; however, it did not result in differences between groups for desired outcomes of either caloric intake or body weight. Future interventions might either deliberately involve caregivers or reduce burden for both patients and caregivers. © 2013 Academy of Nutrition and Dietetics.