The increased reliance on computer systems in the modern world has created a need for engineering reliability control of computer systems to the highest possible standards. This is especially crucial in high-assurance and mission critical systems. Software quality classification models are one of the important tools in achieving high reliability. They can be used to calibrate software metrics-based models to detect fault-prone software modules. Timely use of such models can greatly aid in detecting faults early in the life cycle of the software product. Individual classfiers (models) may be improved by using the combined decision from multiple classifiers. Several algorithms implement this concept and have been investigated. These combined learners provide the software quality modeling community with accurate, robust, and goal oriented models. This paper presents a comprehensive comparative evaluation of three combined learners, Bagging, Boosting, and Logit-Boost. We evaluated these methods with a strong and a weak learner, i.e., C4.5 and Decision Stumps, respectively. Two large-scale case studies of industrial software systems are used in our empirical investigations.