Background Rivaroxaban is a non‐vitamin K oral anticoagulant and has been approved for prevention of stroke and systemic embolism in patients with non‐valvular atrial fibrillation (AF). Current labeling recommends 20 mg once a day (q.d.) as a standard dose and a reduced dose of 15 mg q.d. in patients with renal impairment. Hypothesis The aim of this study was to analyze the adherence to current labeling concerning initial rivaroxaban dosing and to determine whether potential lack of such adherence is medically justified. Methods Patients with AF initiated on rivaroxaban between January 1, 2016 and January 31, 2017, were identified in the Heart Center Leipzig database. Health records were screened to identify prescribed rivaroxaban dose, presence or absence of renal impairment, patient characteristics, further dosing‐relevant diagnoses and co‐medication with antiplatelet drugs and non‐vitamin K oral anticoagulants (NOACs). Results We identified a total of 378 consecutive patients. In 282 cases (74.6%), rivaroxaban was prescribed in a standard dose and in 96 (25.4%) in a reduced dose. Out of 96 patients receiving a reduced dose, 50 (52.1%) did not meet labeling criteria for dose reduction. In uni‐ and multivariate regression analysis, estimated glomerular filtration rate (eGFR) (odds ratio [OR] 0.34, 95% confidence interval [CI] 0.12‐0.95, P = .04) was the only independent predictor of rivaroxaban underdosage. Conclusions In clinical practice, rivaroxaban dosing is frequently incoherent with labeling. In this study, rivaroxaban was often administered underdosed. Potentially inappropriate dose reduction was significantly associated with eGFR, the same factor that is used as criterion for dose adjustment.