Genetically-engineered (GE) crops such as those expressing insecticidal Bacillus thuringiensis (Bt) toxin genes have the potential to greatly reduce the use of broad spectrum insecticides and increase crop productivity. However, development of resistance by the target insect species is an important consideration in the deployment of this strategy. In areas where GE crops are deployed on a large scale, current resistance management strategies rely on a ‘refuge strategy’, comprising the incorporation of a certain proportion of non-GE plants in the agro-ecosystems, to conserve susceptible individuals of the target pests. In the USA, simulation models indicate that at least 20% of the crop should be non-Bt plants. In Africa, the target lepidopteran stemborers attack a wide range of wild grass species as well as cultivated cereal crops. Wild grasses generally occur in the vicinity of maize and other cereal fields, and may provide a refuge if GE crops are in the farming systems. To assess the quality of these grasses as refuges, it is critical to obtain information about their size and spatial distribution. In this study, we have assessed the abundance and diversity of alternative refuge of stemborers, mainly wild grasses occurring in the proximity of maize fields, in Trans-Nzoia district, one of the most important maize growing areas in Kenya. The proportion of wild host plants relative to maize was found to decline from 100% during the non-cropping season to <8% during the maize-growing season. The Shannon-Weaver diversity index indicated high variation in the diversity of wild hosts of stemborers between agro-ecological zones in the district. The results of this study are discussed in light of the possible role that wild host plant species might play in stemborer resistance management following the introduction of Bt maize.