In Muyinga sector, Burundi, an area with good vaccination levels against measles and recent low incidence of measles, a major outbreak of measles in 1988 raised questions about the efficacy of the immunization programme. To help answer these questions, we 1) reviewed programme data on doses of measles vaccine administered, vaccine coverage, and measles incidence, and 2) conducted a census of the affected area to examine vaccine efficacy and measles mortality. We found that between 1980 and 1988 in Burundi, 1) measles vaccine coverage by age 1 had increased from 0% to 55%, 2) the incidence of reported measles cases declined from 12.1/1000 to 6.2/1000, 3) reported measles mortality dropped from 0.18/1000 to 0.08/1000, and 4) the interepidemic period had increased from 25 to 35 months. In the census, the best estimate of measles vaccine efficacy administered at 9 months of age was 73%. Measles increased the risk of death by 2.5-fold with the effect limited to the first month after measles. This outbreak demonstrated the 'post-honeymoon period' epidemic predicted by mathematical models in which outbreaks occur among accumulated susceptibles in a partially immunized population. Understanding this phenomenon is important in providing a basis for improved strategies of measles control. Such outbreaks present new challenges to newly maturing immunization programmes in improving skills in surveillance, outbreak investigation, and public relations.