Most of our knowledge of transcriptional regulation is based on the study of mRNA abundance. The development of high-throughput approaches has revealed pervasive transcription in all genomes that have been investigated so far. This has uncovered a highly interleaved transcriptome organization that involves thousands of coding and non-coding RNAs. Our ability to study transcriptional regulation is further complicated by the complexity within each locus. Each gene expresses diverse transcript isoforms using alternative promoters, exons and terminators. This generates alternative RNA molecules, i.e. isoforms with different length and sequence from the same gene. We have recently developed an approach (transcript isoform sequencing, TIF-Seq) to investigate transcript isoform variation at the genome-wide level. TIF-Seq, by circularizing cDNA molecules and sequencing through the 5'-3' junction, allows to jointly, unambiguously determine the start and end sites of each RNA molecule within a sample. We are further developing this approach to better characterize previously undistinguishable transcripts (e.g., those that arise from neighbouring genes, short abortive transcripts, bicistronic messages...). In addition we plan to expand this approach to the study of mammalian transcriptomes.