The quadratic assignment problem (QAP) is a well-known combinatorial optimization problem with a wide variety of practical applications. Although many heuristics and semi-enumerative procedures for QAP have been proposed, no dominant algorithm has emerged. In this paper, we describe a genetic algorithm (GA) approach to QAP. Genetic algorithms are a class of randomized parallel search heuristics which emulate biological natural selection on a population of feasible solutions. We present computational results which show that this GA approach finds solutions competitive with those of the best previously-known heuristics, and argue that genetic algorithms provide a particularly robust method for QAP and its more complex extensions.