Outside sub-Saharan Africa, Anopheline mosquito exophagic and/or crepuscular behaviour patterns imply that insecticide-treated nets may provide incomplete protection from malaria-infective mosquito bites. Supplementary repellent treatment has been recommended in such circumstances, especially where vectors are exophilic and so are not susceptible to residual insecticide spraying. As maintaining complete usage of repellents in a community is unrealistic, the potential negative impact on non-users of repellent usage by 'neighbours' in the same community needs to be addressed in the context of health policy promoting equity. This study quantifies diversion of host-seeking mosquitoes, from repellent wearing to unprotected individuals, 1 m apart under field conditions in Bolivia. Each of the six volunteer-pairs sat >20 m apart from other pairs. Volunteers were allocated di-ethyl toluamide (DEET) or mineral oil in ethanol control. Treatments were rotated, so that during the trial, both pair-members wore repellent on 72 occasions; both pair-members wore control on 72 occasions; and on 36 occasions, one pair-member wore repellent and the other control. Unprotected (control) pair-members received 36.4% [95% confidence interval (CI): 8.1-72.0%] more Anopheles darlingi landings (P = 0.0096) and 20.4% (95% CI: 0.6-44.0%) more mosquito landings (P = 0.044), when their 'partner' wore repellent than when their partner also wore control. A second, smaller Latin-square trial using 30% lemongrass (Cymbopogon citratus) repellent, with control, obtained 26.0% (95% CI: 5.2-51.0%) more mosquito landings when controls sat with repellent-wearers rather than other controls (P = 0.0159). With incomplete community repellent usage, non-users could be put at an increased risk of malaria. The results also have implications for repellent-efficacy assay design, as protection will appear magnified when mosquitoes are given a choice between repellent-users and non-users.