The objectives of this study were to determine what the Tumbuka people of northern Malawi consider to be the most dangerous childhood diseases, to explore their perceptions about the aetiology, prevention and treatment of these diseases, and to determine how such perceptions have changed over the years. The study was done in Chisinde and surrounding villages in western Rumphi District, northern Malawi. Although a household questionnaire was used to collect some quantitative data, the major data collection methods comprised participant observation, in-depth interviews with mothers with children under five and old men and women, and key informant interviews with traditional healers, traditional birth attendants, village headmen, health surveillance assistants and clinical officers. Informants in this study mentioned chikhoso chamoto, diarrhoea, malaria, measles, and conjunctivitis as the most dangerous childhood diseases in the area. Old men and women added that in the past smallpox was also a dangerous disease that affected both children and adults. Apart from measles and smallpox, community-based health workers and those at the local health centre also mentioned the same list of diseases as the most dangerous diseases prevalent among under-five children. Though health workers and informants mentioned the same diseases, the informants' perspectives about the aetiology and prevention of these diseases and the way they sought treatment during childhood illness episodes, in some cases, differed significantly from those of biomedicine. For example, while health workers said that the signs and symptoms presented by a child suffering from "chikhoso chamoto" were those of either kwashiorkor or marasmus, both young and elderly informants said that a child could contract this illness through contact with a person who had been involved in sexual intercourse. Biomedically, diarrhoea is caused by the ingestion of pathogenic agents, which are transmitted through, among other factors, drinking contaminated water and eating contaminated foods. While young men and women subscribed to this biomedical view, at the same time, just like old men and women, they also believed that if a breastfeeding mother has sexual intercourse, sperms will contaminate her breast milk and, once a child feeds on this milk, he or she will develop diarrhoea. They, in addition, associated diarrhoea with the process of teething and other infections, such as malaria and measles. In malaria-endemic areas such as Malawi, the occurrence of convulsions, splenomegaly and anaemia in children under five may be biomedically attributed to malaria. However, most informants in this study perceived these conditions as separate disease entities caused by, among other factors, witchcraft and the infringement of Tumbuka taboos relating to food, sexual intercourse and funerals. Splenomegaly and convulsions were also perceived as hereditary diseases. Such Tumbuka perceptions about the aetiology of childhood diseases also influenced their ideas about prevention and the seeking of therapy during illness episodes. Apart from measles, other childhood vaccine-preventable diseases (i.e. tetanus, diphtheria, tuberculosis, pertussis and poliomyelitis) were not mentioned, presumably because they are no longer occurring on a significant scale, which is an indication of the success of vaccination programmes. This study reveals that there is no outright rejection of vaccination services in the study area. Some mothers, though, felt pressured to go for vaccination services as they believed that non-vaccinated children were refused biomedical treatment at the local health centres when they fell ill. While young women with children under five mentioned vaccination as a preventative measure against diseases such as measles, they also mentioned other indigenous forms of 'vaccination', which included the adherence to societal taboos, the wearing of amulets, the rubbing of protective medicines into incisions, isolation of children under five (e.g. a newly born child is kept in the house, amongst other things, to protect him or her against people who are ritually considered hot because of sexual intercourse) who are susceptible to disease or those posing a threat to cause disease in children under five. For example, since diarrhoea is perceived to be caused by, among other things, a child feeding on breast milk contaminated with sperms, informants said that there is a strong need for couples to observe postpartum sexual intercourse. A couple with newly delivered twins is isolated from the village because of the belief that children will swell if they came into contact with them. Local methods of disease prevention seem therefore to depend on what is perceived to be the cause of the illness and the decision to adopt specific preventive measures depends on, among other factors, the diagnosis of the cause and of who is vulnerable. The therapy-seeking process is a hierarchical movement within and between aetiologies; at the same time, it is not a random process, but an ordered process of choices in response to negative feedback, and subject to a number of factors, such as the aetiology of the disease, distance, social costs, cost of the therapeutic intervention, availability of medicines, etc. The movement between systems (i.e. from traditional medicine to biomedicine and vice-versa) during illness episodes depends on a number of factors, including previous experiences of significant others (i.e. those close to the patient), perceptions about the chances of getting healed, the decisions of the therapy management group, etc. For example, febrile illness in children under five may be treated using herbs or antipyretics bought from the local grocery shops. When the situation worsens (e.g. accompanied by convulsions), a herbalist will be consulted or the child may be taken to the local health centre. The local health centre refers such cases to the district hospital for treatment. Because of the rapidity with which the condition worsens, informants said that sometimes such children are believed to be bewitched, hence while biomedical treatment is sought, at the same time diviners are also consulted. The therapeutic strategies people resort to during illness episodes are appropriate rational decisions, based on prevailing circumstances, knowledge, resources and outcomes. Boundaries between the different therapeutic options are not rigid, as people move from one form of therapy to another and from one mode of classification to another. Lastly, perceptions about childhood diseases have changed over the years. Old men and women mostly attribute childhood illnesses to the infringement of taboos (e.g. on . sexual intercourse), witchcraft and other supernatural forces. While young men and women also subscribe to these perceptions, they have at the same time also appropriated the biomedical disease explanatory models. These biomedical models were learnt at school, acquired during health education sessions conducted by health workers in the communities as well as during under-five clinics, and health education programmes conducted on the national radio station. Younger people, more frequently than older people, thus move within and between aetiological models in the manner described above.