StringTie is a computational method designed to improve the reconstruction of transcriptomes from RNA-seq reads. It employs a network flow algorithm and optional de novo assembly to assemble complex datasets into transcripts. When compared with other leading transcript assembly programs (Cufflinks, IsoLasso, Scripture, and TraPh), StringTie produces more complete and accurate reconstructions of genes and better estimates of expression levels. On simulated and real data sets, StringTie correctly assembled a significantly higher number of transcripts, with a higher precision and lower false-positive rate. StringTie also runs faster than other assembly software. The method's accuracy is attributed to its optimization criteria, which balance the coverage of each transcript across assemblies, incorporating depth of coverage constraints into the assembly algorithm. StringTie is freely available as open-source software.StringTie is a computational method designed to improve the reconstruction of transcriptomes from RNA-seq reads. It employs a network flow algorithm and optional de novo assembly to assemble complex datasets into transcripts. When compared with other leading transcript assembly programs (Cufflinks, IsoLasso, Scripture, and TraPh), StringTie produces more complete and accurate reconstructions of genes and better estimates of expression levels. On simulated and real data sets, StringTie correctly assembled a significantly higher number of transcripts, with a higher precision and lower false-positive rate. StringTie also runs faster than other assembly software. The method's accuracy is attributed to its optimization criteria, which balance the coverage of each transcript across assemblies, incorporating depth of coverage constraints into the assembly algorithm. StringTie is freely available as open-source software.