A new study presents a set of assessment metrics and visualization techniques for the statistical evaluation of algorithms for quantifying RNA sequencing data. The benchmarks are available as a R/Bioconductor package (http://bioconductor.org/packages/rnaseqcomp). The new set of interpretable assessment metrics relate to the quantification of differential, rather than absolute, expression levels. The authors apply these benchmarks, using two data sets, to seven competing algorithms, including mapping methods such as STAR, TopHat2 and Bowtie2, as well as quantification methods such as Cufflinks, eXpress, Flux Capacitor, kallisto, RSEM, Sailfish and Salmon. Taken together, the benchmarks reveal differences between the algorithms, and the overall performance for the quantification of differential gene expression was poor. An additional webtool (available at http://rafalab.rc.fas.harvard.edu/rnaseqbenchmark) enables users to evaluate the specificity and sensitivity of additional methods.