The objective was to investigate if pre-stroke physical activity is associated with intact cognition early after stroke. The study design was a cross-sectional, register-based study. The study sample included 1111 adults with first stroke (mild or moderate severity) admitted to three Swedish stroke units. The main outcome was cognition. The associations of pre-stroke physical activity, age, sex, smoking, diabetes, atrial fibrillation, previous TIA, statin treatment, hypertension treatment, reperfusion therapies, stroke severity, and education on the outcome cognition were analyzed using binary logistic regression. Physical activity was assessed within 48 h of admittance, and cognition was screened during stroke unit care. The results were: mean age 70 years, 40% women, 61% pre-stroke physically active, and 53% with post-stroke cognitive impairment. Patients with pre-stroke light or moderate physical activity have higher odds for intact cognition compared to inactive: odds ratio (95% confidence interval) 1.32 (0.97-1.80) and 2.04 (1.18-3.53), respectively. In addition to pre-stroke physical activity, people with younger age, a higher level of education, less severe stroke (more mild than moderate), being non-diabetic, and non-smoking have higher odds for intact cognition. In conclusion physical activity before stroke is associated with intact cognition in patients with mild and moderate stroke.