Post-transcriptional processing strongly affects the stability and the relative quantification of RNA molecules, so that steady-state levels of mature RNA, such as mRNAs, rarely reflect accurately the rate of in situ transcription in nuclei by RNA polymerases (RNAPs). The “Global Run-on Sequencing (GRO-Seq)” method, developed in 2008, combines the nuclear run-on assay with next-generation deep sequencing to detect nascent RNA levels to annotate the positions, the relative levels and the orientation of transcriptionally engaged RNA polymerase II (RNAPII) molecules genome-wide. Thus, GRO-Seq is a powerful method to infer mechanistic insights into the multiple levels of transcriptional regulation such as promoter-proximal pausing of RNAP, bidirectional transcription, and enhancer activity. Here, we describe a protocol for mammalian cells that can reliably detect low abundant nascent RNA from both coding and noncoding genomic regions. This protocol can easily be adapted for most mammalian cells to define the transcriptionally active regions of the genome and to measure dynamic transcriptional responses with high sensitivity upon external stimuli.