There has been speculation that the number of cases of diarrhoea-associated haemolytic uraemic syndrome increased abruptly during the early part of the 1980s. to investigate this hypothesis, changepoint models for Poisson variables are applied to two series of data from regional referral units in Newcastle upon Tyne and Birmingham. The series are analysed both separately and jointly, with particular emphasis on determining the number and location of mean changes. The adequacy of the postulated models is considered.