Background: Caregiver burden is commonly experienced in caregivers of adults with cognitive impairment after stroke. This burden can be associated with caregiver-centered factors, including caregiver-perceived relationship quality. Objective: To examine the role of caregiver-perceived relationship quality on caregiver burden at 6 months following stroke. Design: Prospective observational study. Setting: Community settings. Participants: Adults enrolled in two randomized controlled trials after stroke (n = 60) and their caregivers (n = 60). Methods: Three constructs of relationship quality (relationship consensus, cohesion, and satisfaction) were examined as predictors of caregiver burden (the Abbreviated Dyadic Adjustment Scale). A hierarchical logistic regression and a Poisson regression with robust standard errors were used to examine the effect of relationship quality on caregiver burden while controlling for characteristics of stroke survivors and caregivers. Main Outcome Measurements: Caregiver burden was measured with the abridged version of the Zarit Burden Interview. Results: Poor relationship consensus (odds ratio [OR] = 1.48; 95% confidence interval [CI] = 1.06-2.07; P = .02) increased the odds of high caregiver burden after controlling for characteristics of stroke survivors (age, comorbidity, cognitive fluency) and caregivers (gender, self-reported health). Relationship cohesion (OR = 1.03; 95% CI = .84-1.25; P = .81) and relationship satisfaction (OR = 1.53; 95% CI = .75-3.10; P = .24) did not predict caregiver burden. Conclusion: Meaningful and agreeable goals and expectations between dyads are essential to reduce caregiver burden after stroke. Collaborative goal-setting between stroke survivors and their caregivers may be a promising intervention strategy to increase relationship consensus and reduce caregiver burden.