Massively parallel cDNA sequencing, or Total RNA-Seq, has allowed many advances in the characterization and quantification of transcriptomes. This technique offers a new awareness of the multitude of previously unknown coding and non-coding RNA species, particularly small RNAs, including micro RNAs. This work investigates different strategies for Total RNA-Seq using multiple next generation sequencing platforms. Standard RNA-sequencing approaches lead to the loss of information about which strand was present in the original mRNA template. The polarity of the transcript is important for correct annotation of novel genes, identification of antisense transcripts, and for correct determination of gene expression levels in the presence of antisense transcripts. Here, we examine the performance of strand-specific RNA libraries made by direct ligation of adaptor on to the RNA. We analyze the effect of different RNA fragmentation methods and we provide a comparative data analysis. Identification and analysis of small RNA by deep sequencing requires preparation of a di-tagged cDNA library, which leads to contaminating adaptor-dimer formation. We have developed a novel method to generate di-tagged small RNA libraries free of adapter-dimer contamination without introducing any additional enzymatic steps or gel purifications. This method also increases representation of the 2’-O-modified RNAs present in a biological sample. We have also developed a barcode strategy to tag samples during library construction. The multiplexed libraries can then be pooled together before size selection. This technique reduces bias by ligation, increases representation of modified small RNAs and simplifies workflow during library construction for small RNA analysis and discovery.