Two insecticidal bacteria are used as larvicides to control larvae of nuisance and vector mosquitoes in many countries, Bacillus thuringiensis ssp. israelensis and B. sphaericus. Field studies show both are effective, but serious resistance, as high as 50 000-fold, has evolved where B. sphaericus is used against Culex mosquitoes. To improve efficacy and deal with even greater potential problems of resistance, we previously developed several recombinant larvicidal bacteria that combine the best mosquitocidal proteins of these bacteria. In the present study, we report laboratory selection studies using our best recombinant strain against larvae of Culex quinquefasciatus. This recombinant, Bti/BsBin, is a strain of B. thuringiensis ssp. israelensis engineered to produce a large amount of the B. sphaericus binary (Bin) toxin, which makes it more than 10-fold as mosquitocidal as the its parental strains. Here we show that larvae exposed to Bti/BsBin failed to develop significant resistance after 30 successive generations of heavy selection pressure. The highest level of resistance obtained at the LC95 level was 5.2-fold, but declined to less than two-fold at the 35th generation. Testing the selected populations against B. sphaericus alone showed resistance to Bin evolved, but was masked by combination with B. thuringiensis ssp. israelensis. These results suggest that recombinant bacterial strains have improved mosquito and vector management properties compared with the wild-type strains used in current commercial formulations, and should prove useful in controlling important human diseases such as malaria and filariasis on a long-term basis, even when used intensively under field conditions.