IMPORTANCE: Older adults are particularly vulnerable to loneliness and its physical and psychosocial sequelae, but scalable interventions are lacking, especially during disasters such as pandemics. OBJECTIVE: To compare the effects of layperson-delivered, telephone-based behavioral activation and mindfulness interventions vs telephone-based befriending on loneliness among at-risk older adults. DESIGN, SETTING, AND PARTICIPANTS: This assessor-blinded, 3-arm randomized clinical trial screened Chinese older adults through household visits and community referrals from April 1, 2021, to April 30, 2023, in Hong Kong. Eligible participants (≥65 years of age) who were lonely, digitally excluded, living alone, and living below the poverty line and provided consent to participate were randomized into behavioral activation, mindfulness, and befriending groups. Assessments were conducted at baseline, 1 month, and 3 months. INTERVENTION: As part of the Helping Alleviate Loneliness in Hong Kong Older Adults (HEAL-HOA) dual randomized clinical trial, 148 older laypersons were trained to deliver a twice-weekly 30-minute intervention via telephone for 4 weeks. MAIN OUTCOMES AND MEASURES: The primary outcome was loneliness measured by the UCLA Loneliness Scale (range, 20-80) and the De Jong Gierveld Loneliness Scale (range, 0-6), with higher scores on both scales indicating greater loneliness. Secondary outcomes were depression, perceived stress, life satisfaction, psychological well-being, sleep quality, perceived social support, and social network. RESULTS: A total of 1151 participants (mean [SD] age, 76.6 [7.8] years; 843 [73.2%] female) were randomized to the behavioral activation (n = 335), mindfulness (n = 460) or befriending (n = 356) group. Most were widowed or divorced (932 [81.0%]), had primary education or below (782 [67.9%]), and had 3 or more chronic diseases (505 [43.9%]). Following intention-to-treat principles, linear mixed-effects regression model analyses showed that loneliness measured by the UCLA Loneliness Scale was significantly reduced in the behavioral activation group (mean difference [MD], -1.96 [95% CI, -3.16 to -0.77] points; P < .001]) and in the mindfulness group (MD, -1.49 [95% CI, -2.60 to -0.37] points; P = .004) at 3 months compared with befriending. Loneliness measured by the De Jong Gierveld Loneliness Scale was not significantly reduced at 3 months in the behavioral activation group (MD, -0.06 [95% CI, -0.26 to 0.13] points; P > .99]) but was in the mindfulness group (MD, 0.22 [95% CI, 0.03 to 0.40] points; P = .01) at 3 months compared with befriending. In the behavioral activation and mindfulness groups, sleep quality improved compared with befriending, but perceived stress increased. Psychological well-being and perceived social support improved in the behavioral activation group. No statistically significant between-group differences were observed in depression, life satisfaction, or social network. CONCLUSION AND RELEVANCE: In this randomized clinical trial, scalable psychosocial interventions delivered remotely by older laypersons appeared promising in reducing later life loneliness and addressing the pressing mental health challenges faced by aging populations and professional geriatric mental health workforce shortages. Further research should explore ways to maximize the clinical relevance and cost-effectiveness of these interventions. TRIAL REGISTRATION: Chinese Clinical Trial Registry Identifier: ChiCTR2300072909.