RNA sequencing (RNA-seq) is a well-validated tool for detecting gene fusions in fresh frozen tumors; however, RNA-seq is much more challenging to use with formalin-fixed, paraffin-embedded (FFPE) tumor samples. We evaluated the performance of RNA-seq to detect gene fusions in clinical FFPE tumor samples. Our assay identified all 15 spiked-in NTRK fusions from RNA reference material and six known fusions from five cancer cell lines. Limit of detection for the assay was assessed with a series of dilutions of RNA from the cell line H2228. These fusions can be detected when the dilution is down to 10%. Good intra-assay and interassay reproducibility was observed in three specimens. For clinical validation, the assay detected 10 of 12 fusions initially identified by a DNA panel (covering 23 gene fusions) in clinical specimens (83.3% sensitivity), whereas one fusion (MET fusion) was identified in another 34 fusion-negative tumor specimens as determined by the DNA panel (negative prediction value of 94.3%). This MET fusion was confirmed by RT-PCR and Sanger sequencing, which found that this is a false-negative result for the DNA panel. The assay also identified 26 extra fusions not covered by the DNA panel, 20 (76.9%) of which were validated by RT-PCR and Sanger sequencing. Therefore, this RNA assay has reasonable performance and could complement DNA-based next-generation sequencing assays. Copyright © 2020 Association for Molecular Pathology and American Society for Investigative Pathology. Published by Elsevier Inc. All rights reserved.