BACKGROUND: Post-stroke cognitive impairments have been shown to increase the risk of depression, however the mechanisms underpinning this association are not well understood. The theory of social isolation in chronic illness proposes that the relationship between symptom severity in chronic conditions and mood disorders is mediated by social isolation. This study therefore aimed to explore the impact of social isolation on depression in stroke survivors with and without cognitive impairments. METHODS: Stroke survivors were recruited ≥6 months post-stroke from the Oxford Screening Programme and completed assessments of cognitive function, social isolation, and depression. Measures of social isolation evaluated both subjective feelings of loneliness and objective social disconnectedness. Multiple linear regressions examined associations of cognition and social isolation with depression symptoms, and serial mediation analyses assessed potential mediating effects of loneliness and social disconnectedness. RESULTS: Eighty-five participants completed the study measures. Chronic cognitive impairments predicted depressive symptoms in stroke survivors (β = 1.11, p < 0.001). This relationship was mediated by feelings of loneliness, which were associated with higher depression scores in participants with more severe cognitive impairments (indirect effect [IE] = 0.294, p < 0.05). Whilst significant direct effects were observed between all variables in the mediation analyses (all p < 0.05), there was no evidence for indirect effects of social disconnectedness. CONCLUSIONS: Subjective experiences of loneliness, but not objective social disconnectedness, may increase the risk of depression in stroke survivors with cognitive deficits. These findings suggest that feelings of loneliness may be a suitable target for intervention in post-stroke depression.