The effect of bednet use was investigated, without undertaking a specific intervention, in four cross-sectional community-based surveys in 10 villages of a highly endemic area of Papua New Guinea. Over half (55%) of the villagers interviewed reported that they had used a bednet on the previous night. In general and after adjustment for age, village and housing characteristics, bednet users, particularly children, had lower parasite prevalences and spleen rates and less enlarged spleens than non-users. However, users were similar to non-users in terms of fever reported for the previous week, axillary temperature, parasite density and haemoglobin level. The prevalence of antibody to the ring erythrocyte surface antigen and the major merozoite surface antigen 2 was lower in users than non-users. The association with malariometric indices and immune responses remained significant when bednet users were compared with non-users in houses without bednets. Thus, untreated bednets do not reduce malaria transmission sufficiently to decrease morbidity. They might paradoxically increase the risk of clinical malaria by lowering the development of humoral immunity.