Virulence factors of pathogenic bacteria (adhesins, toxins, invasins, protein secretion systems, iron uptake systems, and others) may be encoded by particular regions of the prokaryotic genome termed pathogenicity islands. Pathogenicity islands were first described in human pathogens of the species Escherichia coli, but have recently been found in the genomes of various pathogens of humans, animals, and plants. Pathogenicity islands comprise large genomic regions [10-200 kilobases (kb) in size] that are present on the genomes of pathogenic strains but absent from the genomes of nonpathogenic members of the same or related species. The finding that the G+C content of pathogenicity islands often differs from that of the rest of the genome, the presence of direct repeats at their ends, the association of pathogenicity islands with transfer RNA genes, the presence of integrase determinants and other mobility loci, and their genetic instability argue for the generation of pathogenicity islands by horizontal gene transfer, a process that is well known to contribute to microbial evolution. In this article we review these and other aspects of pathogenicity islands and discuss the concept that they represent a subclass of genomic islands. Genomic islands are present in the majority of genomes of pathogenic as well as nonpathogenic bacteria and may encode accessory functions which have been previously spread among bacterial populations.