BackgroundLoneliness is a risk factor for numerous negative life outcomes and diseases among older adults. Physical exercise can reduce social isolation among older adults and improve their health status. This study focused on the relationships among physical exercise, perceived social support, resilience and loneliness to provide theoretical support for physical exercise interventions to address loneliness in older adults.MethodsThis study used a structural equation model and bootstrap method to test a serial mediation model of physical exercise and loneliness and the role of perceived social support and resilience. In this descriptive and cross-sectional study, a questionnaire survey was conducted on 516 older adults from 5 January 2024 to 20 January 2024. The data were collected using the physical activity rating scale (PARS-3), UCLA loneliness Scale-8 (ULS-8), perceived social support scale (PSSS) and Connor-Davidson Resilience Scale (CD-RISC).ResultsThe results showed that physical exercise among older adults had a direct negative relationship with loneliness (Effect = -0.049, 95% CI: -0.072 to -0.027). In addition, perceived social support and resilience mediated the relationship between physical exercise and loneliness both independent and serially, with the independent mediation effect of perceived social support being − 0.041 (95% CI: -0.056 to -0.028), the independent mediation effect of resilience being − 0.018 (95% CI: -0.026 to -0.011), and the serial mediation effect of perceived social support and resilience being − 0.009 (95% CI: -0.015 to -0.005). The total indirect effect of the three mediation paths was 58.47%.Conclusions(1) In this study, physical exercise and loneliness in older adults were negatively correlated. Perceived social support and resilience play mediating roles in the relationship between physical exercise and loneliness. (2) This study suggests that participation in group events or team sports is valuable for alleviating loneliness caused by a lack of social interaction.