A model is designed and implemented to construct a 'flyable,' least- risk route for strike aircraft from takeoff to target, through enemy radars, in a defined area of operations. A network is fust constructed by discretizing the airspace into a three-dimensional grid of nodes and then connecting adjacent nodes with arcs. A shortest-path model in this network is then constructed with arc lengths that are a function of the probability of detection by radars monitoring the area of operations. A side constraint on fuel consumption ensures that routes are feasible. Lagrangian relaxation is used to incorporate this constraint into the problem and a shortest-path algorithm solves a sequence of shortest-path sub-problems to obtain a near-optimal route. AROMA (Automatic Route Optimization Model for Aircraft) is implemented in C++ on a Silicon Graphics Onyx computer with 192 megabytes of memory. Test problems comprising 240,000 nodes and more than 2 million arcs are used to evaluate the model. Realistic routes are generated in approximately 2 to 3 minutes. A graphical interface displays the routes and facilitates interactive analysis and model evaluation.