In order reduction, complex systems are approximated by comparatively lower order models by retaining the inherent features of the original systems. The main objective is to design a low order controller with lesser number of states to control the higher order system effectively. In this work, the well established balanced realization technique, with frequency weights, is applied for the order reduction of continuous time uncertain systems. The proposed technique results in a stable reduced order model corresponding to higher order system. Test examples available in the literature are considered to demonstrate the results of the proposed technique and a comparison with other methods for order reduction of uncertain systems is given.