Influx of refugees and establishment of camps or settlements in malaria endemic areas can affect the distribution and burden of malaria in the host country. Within a decade of the Soviet invasion of Afghanistan and the arrival of 2.3 million Afghan refugees in Pakistan's North West Frontier Province, the annual burden of malaria among refugees had risen ten fold from 11,200 cases in 1981 to 118,000 cases in 1991, a burden greater than the one reported by the Pakistan Ministry of Health for the entire Pakistani population. Political developments in the 1990s led to over half the refugee population repatriating to Afghanistan, and the Afghan Refugee Health Programme (ARHP) was scaled down proportionately. Districts in which the ARHP recorded a reduced incidence of malaria began to show an increased incidence in the statistics of the Pakistan government health programme. This and other evidence pointed to a change in health seeking practices of the refugees who remained in Pakistan, with many turning from ARHP to Pakistani health services as aid declined. Comparison of the two sources of data produced no evidence for the spatial distribution of malaria in NWFP having changed during the 1990s. Nor was there any evidence for the presence of refugees having increased the malaria burden in the Pakistani population, as is sometimes alleged. This highlights the risk of misinterpreting health trends when parallel health services are operating. Over the decade incidence in the refugee camps decreased by 25% as a result of control activities, and by 1997 the burden among remaining refugees had fallen to 26,856 cases per annum. These trends indicate that the burden would continue to fall if political conditions in Afghanistan were to improve and more refugees returned to their homeland.