Resistance of the dengue vector to temephos stimulated its substitution for Bacillus thuringiensis var. israelensis (Bti) since 2001 in Brazil. The persistence of the two Bti formulations employed at that time by the Health Ministry, Vectobac G and Aquabac G, was assayed under laboratory and outdoor conditions. Both formulations were tested at 0.2 g/10 liters of water, the same concentration applied in the field for vector control. The tests were done against Ae. aegypti third instar larvae (Rockefeller strain). In the laboratory, Vectobac G and Aquabac G caused at least 95% mortality until 101 and 45 days after treatment, respectively. In the outdoor assays, test containers of different materials were treated with either formulation and placed in a shaded area. Larvae were introduced each 3-6 days and mortality was recorded 24 and 48 hours later. In the first set of assays, performed in June 2001, mortality levels of 70% or more were attained for 2-5 weeks for both formulations in all containers. The exception was for the iron one that rusted, resulting in low mortality after seven days. In the second set of assays (August 2001), 70% mortality was attained for just 1-2 weeks for all the containers and both formulations.