Transcriptome analysis by deep sequencing, more commonly known as RNA-seq is, becoming the method of choice for gene discovery and quantitative splicing detection. We published a double-random priming RNA-seq approach capable of generating strand-specific information [Li et al., Proc Natl Acad Sci USA 105:20179-20184, 2008]. Poly(A)+ RNA from a treated and an untreated sample were utilized to generate RNA-seq libraries that were sequenced on the Illumina GA1 analyzer. Statistical analysis of approximately ten million sequence reads generated from both control and treated cells suggests that this tag density is sufficient for quantitative analysis of gene expression. We were also able to detect a large fraction of reads corresponding to annotated alternative exons, with a subset of the reads matching known and detecting new splice junctions. In this chapter, we provide a detailed, bench-ready protocol for the double-random priming method and provide user-friendly templates for the curve-fitting model described in the paper to estimate the tag density needed for optimal detection of regulated gene expression and alternative splicing.