The air traffic service (ATS) telegraphs are composed of fixed information groups. The transmission errors and artificial factors lead to the nonstandard contents of telegraphs, which make the difficulties for the computer to process. In order to solve the problem, pattern recognition and neural network, which can increase the rate of recognition and rate of picking-up right information greatly, are adopted to process the nonstandard messages. This is the key for intelligentized air traffic control. Numerical simulation shows that this method has higher recognition rate and more efficiency. The other excellences of this method are easy to maintain and update.