Clinical scores (i.e. CHA2DS2-VASc) are the mainstay of thromboembolic risk management in nonvalvular atrial fibrillation. Nonetheless, they bear some limitations to precisely define risk-benefit ratio of oral anticoagulation (OAC), both with vitamin K antagonists and with novel direct oral anticoagulants, especially in patients with low-intermediate scores. Cardiovascular imaging, allowing directly visualization of those pathophysiological alterations, which may lead to the formation of intracardiac thrombi, offers itself as a unique tool helping to refine thromboembolic risk stratification. Many parameters have been tested, focusing primarily on functional and morphological variables of the left atrium and left atrial appendage (LAA). Left atrium volume and LAA peak flow velocity have, for a longtime, been associated with increased thromboembolic risk, whereas some new parameters, such as left atrium fibrosis assessed by late-gadolinium enhanced (LGE) MRI, left atrium and LAA strain and LAA morphology have more recently shown some ability in predicting embolic events in atrial fibrillation patients. Overall, however, these parameters have seen, to date, scarce clinical implementation, especially because of the inconsistency of validated cutoffs and/or strong clinical evidence driven by technical limitations, such as expensiveness of the technologies (i.e. MRI or computed tomography), invasiveness (i.e. transesophageal echocardiography) or limited reproducibility (i.e. LGE MRI). In conclusion, to date, cardiovascular imaging plays a limited role; however, validation and diffusion of the new techniques hereby systematically presented hold the potential to refine thromboembolic risk stratification in nonvalvular atrial fibrillation.