Haemophilus influenzae lipopolysaccharide1, 2. We would like to stress not merely the generation of diversity, but also the rate at which it is generated. It is fascinating to analyse the relative extent to which different pathogenic bacteria invest in gene regulation and gene variation to adapt to host environments. Escherichia coli, with its comparatively large genome, appears to have made a major investment in classical gene regulation using an abundance of twocomponent sensor transducers and transcriptional regulators to modulate its phenotype in response to changes in its environment. A distinctive and contrasting feature of H. influenzae is its use of hypermutation through slippage of simple repeats (microsatellites), within or adjacent to a sub-set of genes, to diversify its population, as discussed by Weiser. It is our theory that these hypermutable sequences have evolved because H. influenzae has a limited number of generations in which to generate enough genetic diversity to survive in a rapidly fluctuating environment. The details of this theory are explained below. Transmission between hosts, critical to the basic reproductive ratio3, occurs by the faecal–oral route in the case of E. coli, whereas for H. influenzae, person-to-person spread occurs through transfer of respiratory secretions. In general, faecal–oral spread allows the transfer of large populations of organisms that have the potential to contain many different genotypes. By contrast, respiratory spread is comparatively inefficient in that relatively few of the expelled organisms reach the tissues of the potential recipient host. The acquisition of H. influenzae by a new host might therefore be mediated by a substantially smaller number of founder organisms. For obligate commensals or pathogens, such as H. influenzae, the challenges of co-existence with humans are very stringent and the metaphors of the ‘gene for gene arms race’4 and the ‘Red Queen’5 are especially apt. Unlike E. coli, H. influenzae has no respite, for example through sojourning in the non-evolving and less demanding environment of inanimate reservoirs. H. influenzae is incessantly under threat of extinction if it cannot adapt to the polymorphisms that distinguish genetically distinct hosts, the multifarious landscape of the respiratory tract and the extensive repertoire of innate and acquired host clearance mechanisms. For organisms existing in such rapidly fluctuating environments, survival is a race in which a critical