A study was performed in Quelimane City, Mozambique, using the WHO EPI cluster sampling methodology to assess its applicability for evaluating the impact of the programme on target disease incidence rates in urban areas. Though the methodology was straightforward to use, difficulties were experienced in the retrospective diagnosis of the target diseases, which were felt to limit the reliability of the data obtained. Though results of measles incidence and mortality rates were useful to compare with routine data and discuss with local health staff, the variation in measles incidence rates between epidemic and non-epidemic years may make repeated surveys of this type inappropriate for evaluating programme impact, as it would be difficult to attribute any reduction found in one particular year to immunization and not to the variation in disease incidence which is known to occur in the absence of vaccination. It is suggested that resources would be more profitably used in establishing sentinel sites and improving the quality and utilization of routine data than in conducting repeated sample surveys.