The RAST Server is a fully automated system for annotating bacterial and archaeal genomes. It identifies protein-encoding, rRNA, and tRNA genes, assigns functions to genes, predicts subsystems present in the genome, and reconstructs the metabolic network. The server provides annotated genomes for download and allows users to browse and compare genomes with those in the SEED environment. The service typically produces annotations within 12–24 hours of submission, with a throughput of 50–100 genomes per day. The quality of the service is judged by accuracy, consistency, and completeness of annotations. RAST uses a growing library of manually curated subsystems and protein families derived from these subsystems (FIGfams) to achieve accuracy. It distinguishes between subsystem-based and non-subsystem-based annotations, using the more reliable subsystem-based assertions for metabolic reconstruction. The server has been used by over 120 external users to annotate over 350 genomes. The RAST server employs a two-step process for annotation: first, expert assertions are created based on literature and extrapolation, then bioinformatics tools project these assertions to new genomes. The server uses existing tools to identify tRNA and rRNA genes, then uses GLIMMER2 to call protein-encoding genes. It then searches for FIGfams in the genome to determine gene functions and metabolic reconstruction. The server also provides a genome viewer for browsing and comparing annotated genomes. It processes remaining genes against the entire FIGfam collection, and cleans up gene calls by resolving overlaps and adjusting starting positions. Final function assignments are made, and an initial metabolic reconstruction is formed by connecting genes to functional roles in subsystems. The RAST server is freely available for prokaryotic genome annotation and provides results in various formats. It has been used to annotate over 350 genomes and has a throughput of 50–100 genomes per day. The server's annotations are compared to those from KAAS, showing a high degree of similarity. The server is capable of processing hundreds of closely related pathogen genomes and is designed to support analysis of short DNA fragments and metagenomic data. It is also being expanded to include more types of genes and RNA genes. The server is freely available at http://RAST.nmpdr.org.The RAST Server is a fully automated system for annotating bacterial and archaeal genomes. It identifies protein-encoding, rRNA, and tRNA genes, assigns functions to genes, predicts subsystems present in the genome, and reconstructs the metabolic network. The server provides annotated genomes for download and allows users to browse and compare genomes with those in the SEED environment. The service typically produces annotations within 12–24 hours of submission, with a throughput of 50–100 genomes per day. The quality of the service is judged by accuracy, consistency, and completeness of annotations. RAST uses a growing library of manually curated subsystems and protein families derived from these subsystems (FIGfams) to achieve accuracy. It distinguishes between subsystem-based and non-subsystem-based annotations, using the more reliable subsystem-based assertions for metabolic reconstruction. The server has been used by over 120 external users to annotate over 350 genomes. The RAST server employs a two-step process for annotation: first, expert assertions are created based on literature and extrapolation, then bioinformatics tools project these assertions to new genomes. The server uses existing tools to identify tRNA and rRNA genes, then uses GLIMMER2 to call protein-encoding genes. It then searches for FIGfams in the genome to determine gene functions and metabolic reconstruction. The server also provides a genome viewer for browsing and comparing annotated genomes. It processes remaining genes against the entire FIGfam collection, and cleans up gene calls by resolving overlaps and adjusting starting positions. Final function assignments are made, and an initial metabolic reconstruction is formed by connecting genes to functional roles in subsystems. The RAST server is freely available for prokaryotic genome annotation and provides results in various formats. It has been used to annotate over 350 genomes and has a throughput of 50–100 genomes per day. The server's annotations are compared to those from KAAS, showing a high degree of similarity. The server is capable of processing hundreds of closely related pathogen genomes and is designed to support analysis of short DNA fragments and metagenomic data. It is also being expanded to include more types of genes and RNA genes. The server is freely available at http://RAST.nmpdr.org.