Importance: Although the importance of recovery-oriented care for people with severe mental illness (SMI) is widely acknowledged, essential elements such as personalization and involvement of significant others are not adequately implemented in practice. Objective: To determine whether using resource groups (RGs) within flexible assertive community treatment (FACT) has favorable effects on empowerment and recovery-related outcomes in people with SMI. Design, Setting, and Participants: This assessor-blind, multisite randomized clinical trial was conducted from September 1, 2017, to September 30, 2020, with follow-up at 9 and 18 months. A total of 158 participants aged 18 to 65 years meeting the criteria for SMI were randomly allocated to FACT plus RG vs FACT as usual (1:1) in 20 FACT teams throughout the Netherlands. Data were analyzed from September 1, 2020, to January 31, 2021. The study was prespecified in the trial protocol and data from the intent-to-treat population were analyzed. Interventions: In the FACT plus RG condition, patients chose members from their informal and formal networks to form an RG that meets quarterly to discuss self-formulated recovery goals. The RG was integrated into the multidisciplinary support provided by the FACT team. In the FACT as-usual condition, empowerment (defined as overcoming powerlessness and gaining control of one's life) and involvement of significant others was also part of the provided care, but without the structure of the RG. Main Outcomes and Measures: The primary outcome was self-reported empowerment, measured with the Netherlands Empowerment List. Results: A total of 158 participants with SMI (median age, 38 [median absolute deviation, 13] years; 93 men [58.9%]) were randomized to FACT plus RG (n = 80) or FACT as usual (n = 78) care. Intention-to-treat analyses showed that randomization to the RG condition was associated with a clinically significant increase in empowerment (Cohen d, 0.54; 95% CI, 0.21-0.86) and improved outcomes with small to medium effect sizes in terms of quality of life (Cohen d, 0.25; 95% CI, -0.07 to 0.56), personal recovery (Cohen d, 0.38; 95% CI, 0.06-0.69), quality of social contact (Cohen d, 0.24; 95% CI, -0.07 to 0.56), disability (Cohen d, 0.29; 95% CI, -0.03 to 0.60), general functioning (Cohen d, 0.30; 95% CI, -0.01 to 0.62), and social functioning (Cohen d, 0.28; 95% CI, -0.04 to 0.59). No differences between conditions were found regarding psychopathological symptoms, attachment, frequency of social contact, and employment. Compared with FACT as usual, participants who stayed with the assigned treatment in the RG condition were more satisfied with treatment at 9 (Cohen d = 0.45; t135 = -2.62; P = .009) and 18 (Cohen d = 0.41; t116 = -2.22; P = .02) months. Conclusions and Relevance: These findings show that working with RGs improves empowerment and other mental health outcomes in people with SMI who receive community-based mental health services. This method of network-oriented care empowers people with SMI within their own environment. Trial Registration: Netherlands Trial Register Identifier: NL6548.