A review of empirical studies and the development of a simple theoretical framework are used to explore the relationship between Haemophilus influenzae type b (Hib) carriage and disease within populations. The models emphasize the distinction between asymptomatic and symptomatic infection. Maximum likelihood methods are used to estimate parameter values of the models and to evaluate whether models of infection and disease are satisfactory. The low incidence of carriage suggests that persistence of infection is only compatible with the absence of acquired immunity to asymptomatic infection. The slight decline in carriage rates amongst adults is compatible with acquired immunity, but could be a consequence of reduced contacts. The low rate of disease observed in adulthood cannot be explained if protection from disease is a product of previous detectable exposure to Hib alone. We estimate an Ro of 3·3 for Hib in developed countries, which suggests that current immunization programmes may eliminate the infection. Analysis of the disease data set suggests the absence of maternal immunity and increased susceptibility to disease in the oldest age classes.