Patents are distributed through hundreds of collections, divided up by general area. A hybrid classifier system thus can be a powerful solution to difficult patent classification problems. In this study, we present a system for classifying patent documents on a hybrid approach by combining multiple text classifiers (Naïve Bayes, KNN and Rocchio). Decisions made by various text classifiers can be combined by voting and sampling mechanisms in the system. A prototype system was developed and tested in a real world task. The results have indicated that the accuracy of the hybrid approach is more stable than that of any of the three individual text classifiers.