Numerous studies have indicated an increased risk for stroke in patients with migraine, especially migraine with aura; however, many studies used self-reported migraine and only a few controlled for familial factors. We aimed to investigate migraine as a risk factor for stroke in a Swedish population-based twin cohort, and whether familial factors contribute to an increased risk. The study population included twins without prior cerebrovascular disease who answered a headache questionnaire during 1998 and 2002 for twins born 1935-58 and during 2005-06 for twins born between 1959 and 1985. Migraine with and without aura and probable migraine was defined by an algorithm mapping on to clinical diagnostic criteria according to the International Classification of Headache Disorders. Stroke diagnoses were obtained from the national patient and cause of death registers. Twins were followed longitudinally, by linkage of national registers, from date of interview until date of first stroke, death, or end of study on 31 Dec 2014. In total, 8635 twins had any migraineous headache, whereof 3553 had migraine with aura and 5082 had non-aura migraineous headache (including migraine without aura and probable migraine), and 44 769 twins had no migraine. During a mean follow-up time of 11.9 years we observed 1297 incident cases of stroke. The Cox proportional hazards model with attained age as underlying time scale was used to estimate hazard ratios with 95% confidence intervals for stroke including ischaemic and haemorrhagic subtypes related to migraine with aura, non-aura migraineous headache, and any migraineous headache. Analyses were adjusted for gender and cardiovascular risk factors. Where appropriate; within-pair analyses were performed to control for confounding by familial factors. The age- and gender-adjusted hazard ratio for stroke related to migraine with aura was 1.27 (95% confidence interval 1.00-1.62), P = 0.05, and 1.07 (95% confidence interval 0.91-1.26), P = 0.39 related to any migraineous headache. Multivariable adjusted analyses showed similar results. When stratified by gender and attained age of =50 or >50 years, the estimated hazard ratio for stroke was higher in twins younger than 50 years and in females; however, non-significant. In the within-pair analysis, the hazard ratio for stroke related to migraine with aura was attenuated [hazard ratio 1.09 (95% confidence interval 0.81-1.46), P = 0.59]. In conclusion, we observed no increased stroke risk related to migraine overall but there was a modestly increased risk for stroke related to migraine with aura, and within-pair analyses suggested that familial factors might contribute to this association.