INTRODUCTION: High coverage of conventional and long-lasting insecticide treated nets (ITNs and LLINs) in parts of E Africa are associated with reductions in local malaria burdens. Shifts in malaria vector species ratio have coincided with the scale-up suggesting that some species are being controlled by ITNs/LLINs better than others. METHODS: Between 2005-2006 six experimental hut trials of ITNs and LLINs were conducted in parallel at two field stations in northeastern Tanzania; the first station was in Lower Moshi Rice Irrigation Zone, an area where An. arabiensis predominates, and the second was in coastal Muheza, where An. gambiae and An. funestus predominate. Five pyrethroids and one carbamate insecticide were evaluated on nets in terms of insecticide-induced mortality, blood-feeding inhibition and exiting rates. RESULTS: In the experimental hut trials mortality of An. arabiensis was consistently lower than that of An. gambiae and An. funestus. The mortality rates in trials with pyrethroid-treated nets ranged from 25-52% for An. arabiensis, 63-88% for An. gambiae s.s. and 53-78% for An. funestus. All pyrethroid-treated nets provided considerable protection for the occupants, despite being deliberately holed, with blood-feeding inhibition (percentage reduction in biting rates) being consistent between species. Veranda exiting rates did not differ between species. Percentage mortality of mosquitoes tested in cone bioassays on netting was similar for An. gambiae and An. arabiensis. CONCLUSIONS: LLINs and ITNs treated with pyrethroids were more effective at killing An. gambiae and An. funestus than An. arabiensis. This could be a major contributing factor to the species shifts observed in East Africa following scale up of LLINs. With continued expansion of LLIN coverage in Africa An. arabiensis is likely to remain responsible for residual malaria transmission, and species shifts might be reported over larger areas. Supplementary control measures to LLINs may be necessary to control this vector species.