MitoCarta2.0 is an updated inventory of mammalian mitochondrial proteins, providing a more accurate and comprehensive list of genes encoding mitochondrial-localized proteins in humans and mice. The inventory was developed using improved gene transcript models, updated literature curation, enhanced homology mapping, and revised data sets. The human MitoCarta2.0 includes 1158 genes, with 918 from the original inventory and 240 new genes, while the mouse version includes 1158 genes, with 967 from the original inventory and 191 new genes. The updated inventory offers a molecular framework for system-level analysis of mammalian mitochondria. The MitoCarta2.0 inventory was constructed by integrating seven complementary data sources, including proteomic data, yeast homology, coexpression, protein domains, targeting signals, endosymbiont ancestry, and induction. These data were combined using a naïve Bayes integration method, which significantly improves the accuracy of mitochondrial localization predictions compared to individual methods. The resulting inventory includes 1158 human and 1158 mouse genes, with a false discovery rate (FDR) threshold of 5%. The MitoCarta2.0 database is based on human and mouse RefSeq proteins and includes evidence of mitochondrial localization, protein expression across 14 mouse tissues, and protein sequences. The inventory is available online at www.broadinstitute.org/pubs/MitoCarta, where users can access the updated mitochondrial gene identifiers, evidence of mitochondrial localization, and protein sequences. Compared to other mitochondrial databases, MitoCarta2.0 provides a more specific inventory of mammalian mitochondrial components. It includes a higher number of genes with strong evidence of mitochondrial localization and offers a more accurate prediction of mitochondrial proteins. The database is also more flexible, allowing users to adjust the FDR threshold to suit their needs. Despite its limitations, such as being static and not incorporating new literature evidence, MitoCarta2.0 remains a valuable resource for studying mitochondrial pathways in health and disease.MitoCarta2.0 is an updated inventory of mammalian mitochondrial proteins, providing a more accurate and comprehensive list of genes encoding mitochondrial-localized proteins in humans and mice. The inventory was developed using improved gene transcript models, updated literature curation, enhanced homology mapping, and revised data sets. The human MitoCarta2.0 includes 1158 genes, with 918 from the original inventory and 240 new genes, while the mouse version includes 1158 genes, with 967 from the original inventory and 191 new genes. The updated inventory offers a molecular framework for system-level analysis of mammalian mitochondria. The MitoCarta2.0 inventory was constructed by integrating seven complementary data sources, including proteomic data, yeast homology, coexpression, protein domains, targeting signals, endosymbiont ancestry, and induction. These data were combined using a naïve Bayes integration method, which significantly improves the accuracy of mitochondrial localization predictions compared to individual methods. The resulting inventory includes 1158 human and 1158 mouse genes, with a false discovery rate (FDR) threshold of 5%. The MitoCarta2.0 database is based on human and mouse RefSeq proteins and includes evidence of mitochondrial localization, protein expression across 14 mouse tissues, and protein sequences. The inventory is available online at www.broadinstitute.org/pubs/MitoCarta, where users can access the updated mitochondrial gene identifiers, evidence of mitochondrial localization, and protein sequences. Compared to other mitochondrial databases, MitoCarta2.0 provides a more specific inventory of mammalian mitochondrial components. It includes a higher number of genes with strong evidence of mitochondrial localization and offers a more accurate prediction of mitochondrial proteins. The database is also more flexible, allowing users to adjust the FDR threshold to suit their needs. Despite its limitations, such as being static and not incorporating new literature evidence, MitoCarta2.0 remains a valuable resource for studying mitochondrial pathways in health and disease.