Background Measles remains a major public health problem in Mozambique despite significant efforts to control the disease. Currently, health authorities base their outbreak control on data from the routine surveillance system while vaccine coverage and efficacy are calculated based on mathematical projections of the target population. The aim of this work was to assess the quality of the measles reporting system during two outbreaks that occurred in Maputo City (1998) and in Manica Province (2002). Methods Retrospectively, we collected data from the routine surveillance system, i.e. register books at health facilities and weekly provincial and national epidemiological reports. To test whether the provinces registered an outbreak, the distribution of measles cases was compared to an endemic level established based on cases reported in previous years. Results There was a significant under-notification of measles cases from the health facilities to the province and national level. Register books, the primary sources of information for the measles surveillance system, were found to be incomplete for two main variables: "age" and "vaccination status". Conclusion The Mozambican surveillance system is based on poor quality records, receives the notification of only a fraction of the total number of measles in the country and may result in failures do detect epidemics. The measles reporting system does not provide the data needed by Expanded Program on Immunisation managers to make evidence-based decisions, nor does it allow in-depth analysis to monitor measles epidemiology in the country. The progress of Mozambique to the next stage of measles elimination will require an improvement of the routine surveillance system and a stronger Health Information System.