Background: Research indicates that social isolation and loneliness increase the risk of mental disorders, but less is known about the distinct contributions of different aspects of isolation. We aimed to distinguish the pathways through which social disconnectedness (eg, small social network, infrequent social interaction) and perceptions of social isolation (eg, loneliness, perceived lack of support) contribute to anxiety and depression symptom severity in community-residing older adults aged 57–85 years at baseline. Methods: We did a longitudinal mediation analysis with data from the National Social Life, Health, and Aging Project (NSHAP). The study included individuals from the USA born between 1920 and 1947. Validated measures on social disconnectedness, perceived isolation, and depression and anxiety symptoms were used. Structural equation modelling was used to construct complete longitudinal path models. Findings: Using data from 3005 adults aged 57–85 years, we identified two significant longitudinal mediation patterns with symptoms of depression, and two with anxiety symptoms. Overall, social disconnectedness predicted higher subsequent perceived isolation (β=0·09; p<0·0001), which in turn predicted higher depression symptoms (β=0·12; p<0·0001) and anxiety symptoms (β=0·12; p<0·0001). The reverse pathways were statistically supported as well, suggesting bi-directional influences. Interpretation: Social network structure and function are strongly intertwined with anxiety and depression symptoms in the general population of older adults. Public health initiatives could reduce perceived isolation by facilitating social network integration and participation in community activities, thereby protecting against the development of affective disorders.