The Epithelial-to-Mesenchymal Transition (EMT) is a hallmark of cancer metastasis and morbidity. EMT is a non-binary process, and cells can be stably arrested en route to EMT in an intermediate hybrid state associated with enhanced tumor aggressiveness and worse patient outcomes. Understand- ing EMT progression in detail will provide fundamental insights into the mechanisms underlying metastasis. Despite increasingly available single-cell RNA sequencing data that enable in-depth analyses of EMT at the single-cell resolution, current inferential approaches are limited to bulk microarray data. There is thus a great need for computational frameworks to systematically infer and predict the timing and distribution of EMT-related states at single-cell resolution. Here, we develop a computational framework for reliable inference and prediction of EMT-related trajectories from single-cell RNA sequencing data. Our model can be utilized across a variety of applications to predict the timing and distribution of EMT from single-cell sequencing data.