RSEM is a user-friendly software package designed for quantifying gene and isoform abundances from RNA-Seq data, particularly useful for de novo transcriptome assemblies without a reference genome. It outputs abundance estimates, 95% credibility intervals, and visualization files, and can simulate RNA-Seq data. RSEM does not require a reference genome, making it suitable for species without sequenced genomes. The software has superior or comparable performance to other quantification methods on simulated and real data sets. It effectively uses ambiguously-mapping reads, showing that large numbers of short single-end reads are best for gene-level abundance estimates, while paired-end reads can improve the accuracy of isoform relative frequencies within genes. RSEM also provides valuable guidance for designing cost-efficient RNA-Seq quantification experiments.RSEM is a user-friendly software package designed for quantifying gene and isoform abundances from RNA-Seq data, particularly useful for de novo transcriptome assemblies without a reference genome. It outputs abundance estimates, 95% credibility intervals, and visualization files, and can simulate RNA-Seq data. RSEM does not require a reference genome, making it suitable for species without sequenced genomes. The software has superior or comparable performance to other quantification methods on simulated and real data sets. It effectively uses ambiguously-mapping reads, showing that large numbers of short single-end reads are best for gene-level abundance estimates, while paired-end reads can improve the accuracy of isoform relative frequencies within genes. RSEM also provides valuable guidance for designing cost-efficient RNA-Seq quantification experiments.