A parallel immune genetic algorithm for adaptive modulation and cognitive radio spectrum assignment was developed, and simulations are conducted to compare the parallel immune genetic algorithm method with simple genetic algorithm and non-adaptive modulation algorithm. The simulation results show that the algorithm has strong global searching capability and fast convergence speed, and it is efficient and superior to the simple genetic algorithm and non-adaptive modulation algorithm in case of limited total transmitting power and presence of notable bit error rate, while the computation complexity is reduced significantly.