Despite high coverage of indoor interventions like insecticide-treated nets, mosquito-borne infections persist, partly because of outdoor-biting, early-biting and insecticide-resistant vectors. Push-pull systems, where mosquitoes are repelled from humans and attracted to nearby lethal targets, may constitute effective complementary interventions. A partially randomized cross-over design was used to test efficacy of push-pull in four experimental huts and four local houses, in an area with high pyrethroid resistance in Tanzania. The push-pull system consisted of 1.1% or 2.2% w/v transfluthrin repellent dispensers and an outdoor lure-and-kill device (odour-baited mosquito landing box). Matching controls were set up without push-pull. Adult male volunteers collected mosquitoes attempting to bite them outdoors, but collections were also done indoors using exit traps in experimental huts and by volunteers in the local houses. The collections were done hourly (1830hrs-0730hrs) and mosquito catches compared between push-pull and controls. An. gambiae s.l. and An. funestus s.l. were assessed by PCR to identify sibling species, and ELISA to detect Plasmodium falciparum and blood meal sources. Push-pull in experimental huts reduced outdoor-biting for An. arabiensis and Mansonia species by 30% and 41.5% respectively. However, the reductions were marginal and insignificant for An. funestus (12.2%; p>0.05) and Culex (5%; p>0.05). Highest protection against all species occurred before 2200hrs. There was no significant difference in number of mosquitoes inside exit traps in huts with or without push-pull. In local households, push-pull significantly reduced indoor and outdoor-biting of An. arabiensis by 48% and 25% respectively, but had no effect on other species. This push-pull system offered modest protection against outdoor-biting An. arabiensis, without increasing indoor mosquito densities. Additional experimentation is required to assess how transfluthrin-based products affect mosquito blood-feeding and mortality in push-pull contexts. This approach, if optimised, could potentially complement existing malaria interventions even in areas with high pyrethroid resistance.