This study was to evaluate the value of multi-directional strain parameters derived from three-dimensional (3D) speckle tracking echocardiography (STE) for predicting left ventricular (LV) remodeling after ST-elevation myocardial infarction (STEMI) treated with primary percutaneous coronary intervention (PCI) compared with that of two-dimensional (2D) global longitudinal strain (GLS). A total of 110 patients (mean age, 54 ± 9 years) after STEMI treated with primary PCI were enrolled in our study. At baseline (within 24 h after PCI), standard 2D echocardiography, 2D STE and 3D STE were performed to acquire the conventional echocardiographic parameters and strain parameters. At 3-month follow-up, standard 2D echocardiography was repeated to all the patients to determine LV remodeling, which was defined as a 20% increase in LV end-diastolic volume. At 3-month follow-up, LV remodeling occurred in 26 patients (24%). Compared with patients without LV remodeling, patients with remodeling had significantly reduced 2D GLS (−12.5 ± 3.2% vs −15.0 ± 3.1%, p < 0.001), 3D GLS (−9.9 ± 2.2% vs −13.1 ± 2.7%, p < 0.001), 3D global area strain (GAS) (−20.3 ± 3.9% vs −23.3 ± 4.8%, p = 0.005) and 3D global radial strain (GRS) (29.0 ± 7.4% vs 34.3 ± 8.5%, p = 0.007) at baseline, but there is no significant difference in 3D global circumferential strain (GCS) (−12.7 ± 2.9% vs −13.0 ± 3.2%, p = 0.822). Separated multivariate analysis shows that 2D GLS, 3D GLS, 3D GAS and 3D GRS all can be independent predictors of LV remodeling. However, receiver-operating characteristic curve analysis showed that the area under the curve of 3D GLS (0.82) for predicting LV remodeling was significantly higher than that of 2D GLS (0.72, p = 0.034), 3D GAS (0.68, p < 0.001) and 3D GRS (0.68, p < 0.001). In patients after STEMI, 2D GLS, 3D GLS, 3D GAS and 3D GRS but not 3D GCS measured after primary PCI are independent predictors of LV remodeling and 3D GLS is the most powerful predictor among them.