Aims Guidelines endorse long-term oral anticoagulation (OAC) for stroke prophylaxis in patients with atrial fibrillation (AF) and additional stroke risk factors. However, diverse estimates of guideline based OAC treatment are reported. The objective was to provide insight into the variability seen in estimates of OAC treatment uptake and treatment gaps in clinical practice. Methods and results A systematic review was conducted following Cochrane methodology, using comprehensive search terms for AF and anticoagulation. MEDLINE and EMBASE databases (January 2006–2016) were searched, and prospective and retrospective observational studies reporting OAC use included. Study data and patient characteristics were extracted, OAC rates pooled by study factors, and meta-regression analysis conducted to identify factors associated with OAC uptake. Of 8861 records screened, 155 eligible datasets from 147 studies were identified [median (25th, 75th percentiles) 2259 (595, 10,240) patients, mean age 73.8 ± 7.1 years]. The majority of studies included AF patients with CHA2DS2-VASc >1 [median proportion (25th, 75th percentile) 0.92 (0.83, 1.00)]. OAC use varied across studies [median proportion (25th, 75th percentile) 51.2% (36.7%, 64.7%)] with high heterogeneity (I2 = 100.0%). In the meta-regression analysis [coefficient, p], prospective study design (0.093, p = 0.008), general practitioner treating physician (0.105, p = 0.001) and data collection after 2013 (0.072, p = 0.070) were associated with higher OAC use. Consecutive enrollment (−0.075, p = 0.008) and Asian region (−0.181, p < 0.001) were associated with lower OAC use. Conclusion Study factors, geographical and treatment settings are associated with estimates of OAC use and are important considerations for researchers and policy makers to appropriately interpret reported anticoagulation treatment rates. Consideration of these factors may help to more effectively measure interventions, and design studies to improve anticoagulation uptake.