The evaluation of measles vaccination programmes has been problematic because the change in actual numbers of infections and susceptibles over time cannot be directly observed. In this paper, we present a method for estimating the time series of number of susceptibles and infections, as well as the critical vaccination coverage in a vaccinated population. The proposed method is applied to data on measles outbreaks in The Netherlands. We show that the results are self-consistent and in line with available independent estimates. A potential application of the proposed method lies in detecting the loss of herd immunity and assessing the risk of major outbreaks in vaccinated populations.