Crops expressing Bacillus thuringiensis (Bt) insecticidal Cry proteins are grown on millions of hectares. Recommendations to delay resistance are based on a high expression/refugia strategy that aims to kill resistant heterozygotes and enable some susceptible insects to survive. Leaf-dip bioassays on F1 crosses of Malaysian populations of diamondback moth (Plutella xylostella (L)) showed that CrylAc resistance was not fully recessive. The survival of ca 50% of heterozygotes on Bt canola (Brassica napus L) leaves expressing low concentrations of CrylAc agreed with a non-fully-recessive model for resistance. Extrapolations based on log dose-logit mortality regressions for heterozygotes using leaf-dip bioassays showed that a relatively high level of expression, of ca 2000 ng CrylAc mg(-1) total leaf protein, would be required to give 90% mortality to heterozygotes. If high enough levels of expression of Bt toxin to kill heterozygotes cannot be achieved and maintained under field conditions, the effectiveness of the high-dose/refugia strategy would be reduced.