Do malaria parasites enhance the attractiveness of humans to the parasite's vector? As such manipulation would have important implications for the epidemiology of the disease, the question has been debated for many years. To investigate the issue in a semi-natural situation, we assayed the attractiveness of 12 groups of three western Kenyan children to the main African malaria vector, the mosquito Anopheles gambiae. In each group, one child was uninfected, one was naturally infected with the asexual (non-infective) stage of Plasmodium falciparum, and one harboured the parasite's gametocytes (the stage transmissible to mosquitoes). The children harbouring gametocytes attracted about twice as many mosquitoes as the two other classes of children. In a second assay of the same children, when the parasites had been cleared with anti-malarial treatment, the attractiveness was similar between the three classes of children. In particular, the children who had previously harboured gametocytes, but had now cleared the parasite, were not more attractive than other children. This ruled out the possibility of a bias due to differential intrinsic attractiveness of the children to mosquitoes and strongly suggests that gametocytes increase the attractiveness of the children.