For many years, array-based methods have been used successfully to interrogate the transcriptome of the cell. These methods rely on quantification of the hybridization intensity of the interrogated RNAs to predesigned probes. Recently emerging methods have adopted massively parallel sequencing technologies - also known as Next Generation Sequencing (NGS) technologies -instead. So-called RNA-seq experiments, in contrast to the array-based methods, do not rely on hybridization probes, and therefore aim at the exhaustive identification and quantification of all the RNAs present in a sample, without the requirement of knowing their sequences a priori. The continuously increasing throughput of the NGS instruments allows the measurement of a wide dynamic range of RNA abundances within a cell; moreover, the sequencing depth required to detect even very rare transcripts should soon be available and affordable. Bioinformatics challenges, however, arise from the very short length of the sequenced reads, which complicates their assignment to their parent transcript sequences due to the redundancy in alternatively transcribed sequences and sequencing errors. In this chapter, attention is focused on computational methods that address these challenges in order to identify and quantify transcripts, variants, and alternative splicing (AS) events from RNA-seq data.