RNA velocity is a method to estimate the future state of individual cells based on the time derivative of gene expression. It is derived from the difference between unspliced and spliced mRNA in single-cell RNA sequencing data. This approach allows for the prediction of cellular dynamics, including differentiation and lineage development, by analyzing the rate and direction of change in the transcriptome. RNA velocity was validated in the neural crest lineage, multiple published datasets, and in human embryonic brain development. It revealed the branching lineage tree of the developing mouse hippocampus and provided insights into transcription kinetics during development. The method relies on the observation that unspliced and spliced mRNA can be distinguished in single-cell RNA sequencing data, allowing for the estimation of RNA velocity. This velocity is a high-dimensional vector that predicts the future state of individual cells on a timescale of hours. The approach was tested on various datasets, including those from the 10x Genomics Chromium, SMART-seq2, inDrop, and other protocols, and showed robustness to variations in model parameters and gene and cell subsampling. RNA velocity was applied to the developing mouse hippocampus, revealing a complex lineage tree with multiple branches. It identified key cell types such as astrocytes, oligodendrocyte precursors, and pyramidal neurons. The method also demonstrated the ability to predict cell fates by tracing the expression manifold and extrapolating future states. In human embryos, RNA velocity was detected in the developing forebrain, showing a strong velocity pattern originating from proliferating progenitor cells. The method was further validated in other contexts, including the analysis of light-induced neuronal activation in the mouse cortex and the differentiation of intestinal epithelium and oligodendrocytes. RNA velocity provided insights into the dynamics of cell cycle processes and the regulation of gene expression during development. It was also used to analyze the degradation rates of RNA across different cell types, revealing tissue-specific alternative splicing and degradation rates. Overall, RNA velocity offers a powerful tool for understanding the dynamics of cellular processes, particularly in the context of development and lineage tracing. It provides a quantitative framework for analyzing the future state of individual cells based on their current gene expression profile. This method has the potential to greatly enhance the analysis of developmental lineages and cellular dynamics, especially in humans.RNA velocity is a method to estimate the future state of individual cells based on the time derivative of gene expression. It is derived from the difference between unspliced and spliced mRNA in single-cell RNA sequencing data. This approach allows for the prediction of cellular dynamics, including differentiation and lineage development, by analyzing the rate and direction of change in the transcriptome. RNA velocity was validated in the neural crest lineage, multiple published datasets, and in human embryonic brain development. It revealed the branching lineage tree of the developing mouse hippocampus and provided insights into transcription kinetics during development. The method relies on the observation that unspliced and spliced mRNA can be distinguished in single-cell RNA sequencing data, allowing for the estimation of RNA velocity. This velocity is a high-dimensional vector that predicts the future state of individual cells on a timescale of hours. The approach was tested on various datasets, including those from the 10x Genomics Chromium, SMART-seq2, inDrop, and other protocols, and showed robustness to variations in model parameters and gene and cell subsampling. RNA velocity was applied to the developing mouse hippocampus, revealing a complex lineage tree with multiple branches. It identified key cell types such as astrocytes, oligodendrocyte precursors, and pyramidal neurons. The method also demonstrated the ability to predict cell fates by tracing the expression manifold and extrapolating future states. In human embryos, RNA velocity was detected in the developing forebrain, showing a strong velocity pattern originating from proliferating progenitor cells. The method was further validated in other contexts, including the analysis of light-induced neuronal activation in the mouse cortex and the differentiation of intestinal epithelium and oligodendrocytes. RNA velocity provided insights into the dynamics of cell cycle processes and the regulation of gene expression during development. It was also used to analyze the degradation rates of RNA across different cell types, revealing tissue-specific alternative splicing and degradation rates. Overall, RNA velocity offers a powerful tool for understanding the dynamics of cellular processes, particularly in the context of development and lineage tracing. It provides a quantitative framework for analyzing the future state of individual cells based on their current gene expression profile. This method has the potential to greatly enhance the analysis of developmental lineages and cellular dynamics, especially in humans.