Background and Purpose— Our study aim was to estimate risk of incident stroke based on levels of hs-cTn (high-sensitivity cardiac troponin), a specific biomarker indicating myocardial injury, in the general population, patients with atrial fibrillation, and patients with previous stroke. Methods— Embase, PubMed, and Web of Science were searched until March 14, 2019 to identify relevant articles. Randomized controlled trials and cohort studies assessing the risk of incident stroke based on hs-cTn were eligible. Pooled adjusted hazard ratios including 95% CI were calculated using a random-effects model due to study heterogeneity per population, coding of hs-cTn (categorical/continuous data), per hs-cTn subunit (T or I), for low risk of bias, and for all-cause and ischemic stroke separately. Results— We included 17 articles with 96 702 participants. In studies conducted in the general population (n=12; 77 780 participants), the pooled adjusted hazard ratio for incident stroke was 1.25 (CI, 1.10–1.40) for high versus low hs-cTn (as defined by included studies) during an average follow-up of 1 to 20 years (median 10). When categorical data were used, this was increased to 1.58 (CI, 1.26–1.90). The results were robust when accounting for stroke classification (all-cause stroke/ischemic stroke), hs-cTn subunit, risk of bias, and coding of hs-cTn. In patients with atrial fibrillation (4 studies; 18 725 participants), the pooled adjusted hazard ratio for incident stroke was 1.95 (CI, 1.29–2.62) for high versus low hs-cTn. Due to lack of data (one study, 197 participants), no meta-analysis could be performed in patients with previous stroke. Conclusions— This meta-analysis suggests that hs-cTn can be regarded as a risk marker for incident stroke, with different effect size in different subgroups. More research about the association between hs-cTn and incident stroke in high-risk populations is needed, especially in patients with history of ischemic stroke.