Short open reading frames (sORFs) are abundant within eukaryotic transcripts and transcript regions that were previously thought to be non-coding, and evidence for their functionality has been increasing; however, the large-scale identification of functional sORFs has so far proved challenging. Using a new computational approach based on the presence of specific amino-acid-based evolutionary conservation features, Mackowiak et al. have identified over 2,000 novel sORFs in 5 animal species. The sORFs were located in transcripts that had been annotated as non-coding or in the presumed untranslated regions of canonical mRNAs, and some were conserved over long evolutionary distances. Mining of experimental data sets revealed that the newly identified sORFs are expressed and are often translated. The highly annotated results of this study represent a useful resource for further experimental functional analysis of sORFs.