Background We previously constructed a Self-Mutual-Group health management model, focusing on improving individuals' awareness of health management and ability in self-management, mutual-management, and group-management by increasing their self-efficacy. We aimed to evaluate the effectiveness of the Self-Mutual-Group model on social support among the empty-nest elderly (ie, those who live on their own), and to explore the related factors that might have a role in the intervention effect, thus providing information for improving the social support of empty-nest elderly. Methods We did an open-label, randomised controlled trial of six communities in Taiyuan, Shanxi province, China. We included empty-nest elderly individuals aged 60–88 years if they had no cognitive disorder or other mental illnesses, were willing and able to complete the investigation, and were dwelling in the community for at least 1 year before the study. We randomly assigned participants using a computer system to either the intervention group with a 7-month intervention or the control group with no intervention. The 7-month intervention consisted of three stages (self-management at months 1–2, mutual management at months 3–4, and group management at months 5–7) and was held at local health-care or residential activity centres by researchers and community workers assisting in the Self-Mutual-Group-based interventions. We assessed the social support of the participating elderly using the Social Support Rating Scale (SSRS) at baseline and 7 months' follow-up. This study was approved by the institutional review boards of Shanxi Medical University. We obtained informed consent from all participants. This trial is registered with the Chinese Clinical Trial Register, number ChiCTR1800015884 Findings Between Aug 1, 2016, and Sept 30, 2016, we enrolled 350 empty-nest elderly individuals; 167 in the intervention group and 183 in the control group. At 7 months' follow-up, we observed a significant increase between the intervention group and the control group in the mean scores of SSRS, objective support, and subjective support. The main influencing factors of the intervention effect were sex, marriage status, income sources, monthly income and employment status, relationship with children, and self-care ability of daily living. Interpretation The Self-Mutual-Group model was considered a positive intervention and was effective in increasing social support among the empty-nest elderly. Among the empty-nest elderly, men without spouses have a lower socioeconomic status than those with spouses. Therefore, those who do not have the ability for self-care should be the focus of future research and model popularisation to increase their social support effectiveness and improve their quality of life. Furthermore, future research might extend the intervention period and introduce multi-point measurement to better evaluate the intervention process and outcomes. Funding National Natural Science Foundation of China (71874104 and 71403155), Soft Science Research Program in Shanxi (2017041039-2), Fund Program for the Scientific Activities of Selected Returned Overseas Professional in Shanxi Province (2017-16), and Research Project Supported by Shanxi Scholarship Council of China (2017-062).