Due to the increasing complexity in modern radar systems, it has become apparent that new ways of classify received signals are required. In this paper, the performance of a proposed method for radar system identification using hidden Markov model (HMM) is evaluated. In this method, a given radar system is modeled as a finite state automaton. In so, doing, it is possible to uncover the underlying system processes in a probabilistic fashion using hidden Markov models. Artificial deterministic signals are used to show that HMMs can provide adequate signal recognition. It will be shown that a perfect exists that is directly related to the number of symbols per period. A slightly adjusted version of this perfect model provides low false recognition rates in a threat library consisting of three models so long as the observation errors do not become too frequent. The simulation results show that if the observation errors are introduced, then, it becomes necessary to retrain the HMM with an error corrupted version of the original training sequence in order to improve the model's robustness.