Background: Non-valvular atrial fibrillation (NVAF) is a highly prevalent arrhythmia where loss of synchronized atrial contraction increases the risk of intracardiac thrombus particularly within the left atrial appendage (LAA). Anticoagulation is the mainstay of stroke prevention based on the CHA2DS2-VASc score; however, it does not account for LAA structural characteristics. Methods: The research comprises a retrospective matched case-control study of 196 subjects with NVAF who underwent transesophageal echo (TEE). The control group, without thrombus (n=117), was selected from two different groups, both pools had: NVAF and CHA2DS2-VASc score ≥ 3. One group underwent screening TEE prior to Watchman closure device placement from January 2015 to December 2019 (n=74) the second underwent TEE prior to cardioversion from February to October 2014 (n=43). The study group, with thrombus (n=79), included patients with NVAF, TEE study performed between February 2014 and December 2020, and LAA thrombus. The propensity score method was utilized to determine the matched controls while accounting for confounding from prognostic variables resulting in 61 matched pairs included in the analysis data set. LAA ostial area (OA) (calculated from orthogonal measurements 0, 90° or 45, 135°), LAA maximal depth, and peak LAA outflow velocity were measured. Results: Patient characteristics and TEE data were collected (Table [I](#tbl-cap-0001)) and compared using the t-test or chi-square analysis. We observed a lower LAA peak exit velocity in the thrombus group as compared to the control group. Additionally, we found that patients in the thrombus group had smaller LAA OA at 0 and 90 degrees, at 45 and 135 degrees, using largest diameter, as well as using aggregate OA, and smaller maximum LAA depth compared to patients in the control group. Candidate conditional logistic regression models for the outcome of presence of thrombus were evaluated (Table [II](#tbl-cap-0002)). Statistical results from the best-fitting conditional regression model were calculated (Table [III](#tbl-cap-0003)) showing a significant association between aggregate OA and LAA exit velocity with presence of thrombus. Conclusion: Utilizing LAA structural characteristics to predict thrombus formation may help refine current cardioembolic stroke (CES) risk estimation.