Kraken 2 is a memory-efficient metagenomic classification tool that improves upon Kraken 1 by reducing memory usage by 85%, increasing speed fivefold, and introducing a translated search mode for better viral metagenomics analysis. Kraken 2 uses a probabilistic hash table to map minimizers to lowest common ancestor (LCA) taxa, reducing memory usage and improving speed. It also uses a spaced seed searching scheme and a translated search mode for amino acid space. Kraken 2's memory requirement is 15% of Kraken 1's and only 2.5 times that of Centrifuge, the least memory-intensive classifier. Kraken 2's accuracy is comparable to Kraken 1 and other tools, with Kraken 2X providing similar accuracy to Kaiju but using less memory and runtime. Kraken 2 is compatible with Bracken for species-level quantification. Kraken 2's algorithms also allow for further reduction in database size and faster processing of sequencing data. Kraken 2's performance was evaluated on 50 simulated genomes, showing similar or superior accuracy to other nucleotide classifiers and better sensitivity in viral datasets. Kraken 2's memory usage and speed are significantly lower than Kraken 1, with Kraken 2 processing over 93 million reads per minute using 16 threads. Kraken 2's memory usage is also 85% smaller than Kraken 1's. Kraken 2's translated search mode (Kraken 2X) provides improved sensitivity on viral datasets compared to nucleotide-based search. Kraken 2's performance was also evaluated on real sequencing data from the FDA-ARGOS project, showing similar genus-level concordance to other nucleotide classifiers. Kraken 2's accuracy and performance were further validated using strain-exclusion experiments, where it outperformed Kraken 1 in species-level abundance estimation. Kraken 2's use of a probabilistic hash table and minimizer-based subsampling allows for more efficient processing of sequencing data and reduces memory usage. Kraken 2's algorithms also allow for further reduction in database size and faster processing of sequencing data. Kraken 2's performance was evaluated on a variety of datasets, showing its effectiveness in metagenomic classification.Kraken 2 is a memory-efficient metagenomic classification tool that improves upon Kraken 1 by reducing memory usage by 85%, increasing speed fivefold, and introducing a translated search mode for better viral metagenomics analysis. Kraken 2 uses a probabilistic hash table to map minimizers to lowest common ancestor (LCA) taxa, reducing memory usage and improving speed. It also uses a spaced seed searching scheme and a translated search mode for amino acid space. Kraken 2's memory requirement is 15% of Kraken 1's and only 2.5 times that of Centrifuge, the least memory-intensive classifier. Kraken 2's accuracy is comparable to Kraken 1 and other tools, with Kraken 2X providing similar accuracy to Kaiju but using less memory and runtime. Kraken 2 is compatible with Bracken for species-level quantification. Kraken 2's algorithms also allow for further reduction in database size and faster processing of sequencing data. Kraken 2's performance was evaluated on 50 simulated genomes, showing similar or superior accuracy to other nucleotide classifiers and better sensitivity in viral datasets. Kraken 2's memory usage and speed are significantly lower than Kraken 1, with Kraken 2 processing over 93 million reads per minute using 16 threads. Kraken 2's memory usage is also 85% smaller than Kraken 1's. Kraken 2's translated search mode (Kraken 2X) provides improved sensitivity on viral datasets compared to nucleotide-based search. Kraken 2's performance was also evaluated on real sequencing data from the FDA-ARGOS project, showing similar genus-level concordance to other nucleotide classifiers. Kraken 2's accuracy and performance were further validated using strain-exclusion experiments, where it outperformed Kraken 1 in species-level abundance estimation. Kraken 2's use of a probabilistic hash table and minimizer-based subsampling allows for more efficient processing of sequencing data and reduces memory usage. Kraken 2's algorithms also allow for further reduction in database size and faster processing of sequencing data. Kraken 2's performance was evaluated on a variety of datasets, showing its effectiveness in metagenomic classification.