Purpose: The susceptibility vessel sign (SVS) has been described on gradient echo (GRE) magnetic resonance imaging (MRI) in acute ischemic stroke patients by large vessel occlusion. The presence of SVS (SVS+) was associated with treatment outcome and stroke etiology with conflicting results. Based on multicenter data from the THRombectomie des Artères CErebrales (THRACE) study, we aimed to determine if the association between SVS and cardioembolic etiology (CE) was independent of GRE sequence parameters. Material and methods: Patients with a pretreatment brain GRE sequence were identified. Logistic regression tested the association between SVS+, CE, time from onset to imaging and GRE sequence parameters (e.g. echo time, voxel size, field strength). We calculated the sensitivity, specificity, positive and negative predictive values (PPV and NPV) for the SVS to predict a stroke from a CE. Results: An SVS+ was observed in 237 out of 287 (83%) patients. In the univariate analysis, there was a significant association between SVS+ and a CE with an odds ratio (OR) and 95% confidence interval (95% CI) of 2.10 (1.02-4.29), respectively (p = 0.04) but not with GRE sequence parameters. In multivariate analysis, there was an independent relationship between SVS+ and CE (OR [95% CI]: 2.14 [1.02-4.45], p = 0.04). Sensitivity and specificity of SVS+ to predict CE were 0.89 and 0.21, respectively. The PPV and NPV of SVS+ were 0.44 and 0.78, respectively. Conclusion: The presence of SVS is associated to CE, independent of GRE sequence parameters. While the specificity and the PPV of the sign were low, CE seems less likely in the absence of an SVS.