The subject of this thesis is to investigate methods for discrimination between two classes of radar emitters on the basis of the distribution of the radio frequencies. For the first class, the radio frequency is uniformly distributed within a certain frequency band. For the second class, the radio frequency belongs to a set of discrete frequencies where each frequency is equally probable. Two methods has been investigated. The first method is a recursive algorithm that applies multiple Kalman filters. The second method is a clustering algorithm. It is a non-recursive algorithm that for each acquired measurement processes all measurements in a batch. To compare the different methods, the posterior probability of each class was calculated given a sequence of measurements. The posterior probability represents all knowledge of the emitters, and constitutes the base for further decisions. The results show that both methods are capable of discriminating between the two classes. For the method that applies multiple Kalman filters, the probability of each class quickly converges towards the desired results. The number of discrete frequencies is chosen as 10 in the simulations, and the number of required measurements is then approximately 30. For the clustering algorithm, the number of required measurements is substantially larger than for the method with the multiple Kalman filters.