The article introduces RNA velocity, a novel approach to estimate the time derivative of gene expression in single cells, which can predict the future state of individual cells over hours. The method distinguishes unspliced and spliced mRNAs in common single-cell RNA sequencing protocols, allowing for the estimation of RNA velocity without the need for metabolic labeling. The authors validate RNA velocity in the neural crest lineage, demonstrate its utility on multiple datasets and platforms, and reveal the branching lineage tree of the developing mouse hippocampus. They also examine transcription kinetics in human embryonic brain. RNA velocity is expected to aid in the analysis of developmental lineages and cellular dynamics, particularly in humans. The study includes detailed methods for data annotation, analysis, and visualization, and provides software (Velocyto) to implement the computational framework.The article introduces RNA velocity, a novel approach to estimate the time derivative of gene expression in single cells, which can predict the future state of individual cells over hours. The method distinguishes unspliced and spliced mRNAs in common single-cell RNA sequencing protocols, allowing for the estimation of RNA velocity without the need for metabolic labeling. The authors validate RNA velocity in the neural crest lineage, demonstrate its utility on multiple datasets and platforms, and reveal the branching lineage tree of the developing mouse hippocampus. They also examine transcription kinetics in human embryonic brain. RNA velocity is expected to aid in the analysis of developmental lineages and cellular dynamics, particularly in humans. The study includes detailed methods for data annotation, analysis, and visualization, and provides software (Velocyto) to implement the computational framework.