Increased walking among facility-dwelling older adults, who are very old, frail, and have low physical activity, could have substantial health benefits. Multilevel approaches to improving physical activity, based on Ecological Models and Social Cognitive Theory, have not been tested in this population but hold promise for improved effects. This study aimed to investigate the feasibility and outcomes of a 3-month enhanced, multilevel walking intervention, compared to a standard walking intervention, among older adults in retirement communities. Participants in the enhanced intervention group were hypothesized to have improved outcomes compared to those in the standard intervention. Data were collected at baseline (N = 87) and post-intervention (N = 67) from residents in 4 retirement facilities. Sites were quasi-randomized to condition (N = 2 sites per condition). Standard intervention components included pedometers, printed materials, and biweekly group sessions; those in the enhanced intervention also received individual biweekly phone counseling and environmental awareness components. Measures included activity related outcomes (pedometers, sedentary behavior, activities of daily living, on and off-site walking, satisfaction with walking opportunities, neighborhood barriers), physical function, mental health outcomes (quality of life, depression), study satisfaction, and adherence to study components. Data were analyzed using analysis of covariance (ANCOVA) for between group differences and repeated measures ANCOVA for pre-post test changes. None of the outcomes were significantly different between walking intervention conditions except for neighborhood barriers. Standard intervention participants had significantly fewer neighborhood barriers post-intervention compared to the enhanced intervention group. Significant improvements from baseline to post-intervention occurred among the total sample for step counts, neighborhood barriers, walking up stairs, walking off-site, and satisfaction with walking opportunities but significance disappeared after adjustment for covariates. Study satisfaction and adherence was high for both groups. The results of this study suggest that two different types of walking interventions are feasible to conduct and result in improved step counts among facility-dwelling older adults. The most change occurred for environment-related variables. Findings suggest that the context of walking is important for older adults residing in retirement facilities and should be targeted in future interventions. Future studies can build on this novel multilevel approach to improving walking among very old adults.