The cereus group represents sporulating soil bacteria containing pathogenic strains which may cause diarrheic or emetic food poisoning outbreaks. Multiple locus sequence typing revealed a presence in natural samples of these bacteria of about thirty clonal complexes. Application of genomic methods to this group was however biased due to the major interest for representatives closely related to B. anthracis. Albeit the most important food-borne pathogens were not yet defined, existing data indicate that they are scattered all over the phylogenetic tree. The preliminary analysis of the sequences of three genomes discussed in this paper narrows down the gaps in our knowledge of the cereus group. The strain NVH391-98 is a rare but particularly severe food-borne pathogen. Sequencing revealed that the strain must be a representative of a novel bacterial species, for which the name Bacillus cytotoxis is proposed. This strain has a reduced genome size compared to other cereus group strains. Genome analysis revealed absence of sigma B factor and the presence of genes encoding diarrheic Nhe toxin, not detected earlier. The strain B. cereus F837/76 represents a clonal complex close to that of B. anthracis. Including F837/76, three such B. cereus strains had been sequenced. Alignment of genomes suggests that B. anthracis is their common ancestor. Since such strains often emerge from clinical cases, they merit a special attention. The third strain, KBAB4, is a typical psychrotrophe characteristic to unbiased soil communities. Phylogenic studies show that in nature it is the most active group in terms of gene exchange. Genomic sequence revealed high presence of extra-chromosomal genetic material (about 530 kb) that may account for this phenomenon. Genes coding Nhe-like toxin were found on a big plasmid in this strain. This may indicate a potential mechanism of toxicity spread from the psychrotrophic strain community. The results of this genomic work and ecological compartments of different strains incite to consider a necessity of creating prophylactic vaccines against bacteria closely related to NVH391-98 and F837/76. Presumably developing of such vaccines can be based on the properties of non-pathogenic strains such as KBAB4 or ATCC14579 reported here or earlier. By comparing the protein coding genes of strains being sequenced in this project to others we estimate the shared proteome in the cereus group to be 3,000?b200 genes and the total proteome to be 20-25,000 genes.