In direction finding MUSIC (multiple signal classification) is used in different versions. In certain situations second-order algorithms perform well, whereas in other cases higher-order methods should be preferred. That's why the authors propose the combination of these methods to obtain optimal performance in the case of unknown signal properties. In the paper the properties of the different estimators are discussed. Two versions of the combinations are introduced. The most extensive part of the paper deals with the numerical simulation of these two methods. The simulation results show that the combined methods are really working better than the baseline algorithms.