Introduction <p>We evaluated the association between loneliness, social isolation, and stroke risk in individuals with and without diabetes and assessed sex differences in these associations.</p> Research design and methods <p>From the UK Biobank Study, we included 40,684 participants with diabetes (mean age 51.25, 12.5% women, 88.65% white) (self-reported diabetes diagnosis, diabetes medication, HbA1c &gt; 6.5%, or ICD codes) and those without (1:1 matched on sex, race, age, and center). We used Cox regression to examine the association between loneliness, social isolation, and stroke risk and estimated the absolute risk reduction (ARR) of strokes in diabetes and euglycemic groups separately. We also performed Cox regression to evaluate the excess stroke risk associated with diabetes (vs. euglycemic controls) across levels of loneliness and social isolation.</p> Results <p>During a median follow-up of 12 years, within the diabetes cohort, lower levels of loneliness and social isolation were associated with reduced stroke risk (Hazard ratio [HR] for loneliness 0.73, 95% CI 0.62–0.86; social isolation 0.83, 0.72–0.97), and such associations were significantly stronger in men than women (HR for loneliness 0.63, 0.52–0.77 in men, 0.96, 0.72–1.29 in women, p-interaction = 0.02; HR for social isolation 0.79, 0.65–0.95 in men, 0.93, 0.72–1.22 in women, p-interaction = 0.05). ARR related to improved loneliness and social isolation was 1.3% and 0.4% in the diabetes group (equivalent to 13 and 4 stroke cases per 1000 people with diabetes, respectively, during the 12-year period), respectively. As the index of loneliness and social isolation went from highest to lowest, the adjusted stroke risk decreased by 65% and 8%, respectively, in people with diabetes vs. those without.</p> Conclusions <p>Improved loneliness and social isolation may mitigate excess stroke risk in people with diabetes, particularly in men with diabetes.</p>