Abstract Transgenic crops producing Bacillus thuringiensis (Bt) toxins are commonly used for controlling insect pests. Nearby refuges of non-Bt host plants play a central role in delaying the evolution of resistance to Bt toxins by pests. Pervasive fitness costs associated with resistance, which entail lower fitness of resistant than susceptible individuals in refuges, can increase the ability of refuges to delay resistance. Moreover, these costs are affected by environmental factors such as host plant suitability, implying that manipulating refuge plant suitability could improve the success of the refuge strategy. Based on results from a previous study of Trichoplusia ni resistant to Bt sprays, it was proposed that low-suitability host plants could magnify costs. To test this hypothesis, we investigated the association between host plant suitability and fitness costs for 80 observations from 30 cases reported in 18 studies of 8 pest species from 5 countries. Consistent with the hypothesis, the association between plant suitability and fitness cost was negative. With plant suitability scaled to range from 0 (low) to 1 (high), the expected cost was 20.7% with a suitability of 1 and the fitness cost increased 2.5% for each 0.1 decrease in suitability. The most common type of resistance to Bt toxins involves mutations affecting a few types of midgut proteins to which Bt toxins bind to kill insects. A better understanding of how such mutations interact with host plant suitability to generate fitness costs could be useful for enhancing the refuge strategy and sustaining the efficacy of Bt crops.