The different strategies of insecticide resistance management that have been formulated so far consist of delaying the appearance and spread of resistance genes. In this paper, we propose a strategy that can be used even if resistance genes are already present. This strategy consists of applying insecticides in an area smaller than a certain critical size, so that gene flow from the untreated area, combined with the fitness cost of the resistance genes, prevents its frequency reaching high equilibrium value. A two–locus model was analysed numerically to determine population densities at equilibrium as a function of selection coefficients (insecticide selection, fitness costs of resistance genes and dominances), gene flow and size of the treated area. This model indicates that there is an optimal size for the treated area where a minimal and stable density reach equilibrium, and where resistance genes cannot invade. This resistance management strategy seems applicable to a large variety of field situations, but eventually it may encounter obstacles due to a modifier which reduces the fitness costs of resistance genes.