Modeling and experiments have suggested that spatial fuel treatment patterns can influence the movement of large fires. On simple theoretical landscapes consisting of two fuel types (treated and untreated) optimal patterns can be analytically derived that disrupt fire growth efficiently (i.e. with less area treated than random patterns). Although conceptually simple, the application of these theories to actual landscapes is made difficult by heterogeneity (fuels, weather, and topography) compared to the assumptions required for analytical solutions. Here I describe a computational method for heterogeneous landscapes that identifies efficient fuel treatment units and patterns for a selected fire weather scenario. The method requires input of two sets of spatial input data: 1) the current fuel conditions and 2) the potential fuel conditions after a treatment (if it were possible). The contrast in fire spread rate between the two landscapes under the weather scenario conditions indicates where treatments are effective at delaying the growth of fires. Fire growth from the upwind edge of the landscape is then computed using a minimum travel time algorithm. This identifies major fire travel routes (areas needing treatment) and their intersections with the areas where treatments occurred and reduced the spread rate (opportunity for treatment). These zones of treatment 'need and opportunity' are iteratively delineated by contiguous patches of raster cells up to a user-supplied constraint on percentage of land area to be treated. This algorithm is demonstrated for simple and for complex landscapes.