miRBase is a comprehensive resource for microRNA sequences and annotations, providing cataloging, naming, and distribution of microRNA gene sequences. The latest release (v22) includes microRNA sequences from 271 organisms, with 38,589 hairpin precursors and 48,860 mature microRNAs. The database has been updated to improve the quality of annotations and provide more information on the cellular functions of microRNA products. Over 1493 small RNA deep sequencing datasets have been collected, mapping a total of 5.5 billion reads to microRNA sequences, which supports the validity of between 20% and 65% of microRNA annotations in well-studied animal genomes. To enhance functional information, Gene Ontology terms have been annotated against miRBase sequences, and text-mining approaches have been used to search for microRNA gene names in open access articles. Over 500,000 sentences from 18,542 papers contain microRNA names, and these sentences are scored for functional information and linked with 12,519 microRNA entries. The sentences and word clouds provide effective summaries of functional information about specific microRNAs. miRBase is publicly available at http://mirbase.org/.miRBase is a comprehensive resource for microRNA sequences and annotations, providing cataloging, naming, and distribution of microRNA gene sequences. The latest release (v22) includes microRNA sequences from 271 organisms, with 38,589 hairpin precursors and 48,860 mature microRNAs. The database has been updated to improve the quality of annotations and provide more information on the cellular functions of microRNA products. Over 1493 small RNA deep sequencing datasets have been collected, mapping a total of 5.5 billion reads to microRNA sequences, which supports the validity of between 20% and 65% of microRNA annotations in well-studied animal genomes. To enhance functional information, Gene Ontology terms have been annotated against miRBase sequences, and text-mining approaches have been used to search for microRNA gene names in open access articles. Over 500,000 sentences from 18,542 papers contain microRNA names, and these sentences are scored for functional information and linked with 12,519 microRNA entries. The sentences and word clouds provide effective summaries of functional information about specific microRNAs. miRBase is publicly available at http://mirbase.org/.