This paper introduces a technique for simultaneous decomposition of a non-autonomous singularly perturbed system into three subsystems namely slow, fast 1 and fast 2 respectively. Design of a composite controller in terms of the individual subsystem controllers is derived and the decomposition of the optimal control problem of the original high order system into three smaller order optimal control problems, with separate quadratic performance indexes extracted from the quadratic performance index of the original system, is discussed. The proposed technique is applied to the problem of spatial stabilization of Advanced Heavy Water Reactor (AHWR), which is a neutronically large reactor necessitating a control system to suppress xenon induced spatial oscillations. However the AHWR model, with eigenvalues falling in three distinct and widely separated clusters, is ill-conditioned defying even the application of two-time-scale decomposition. In this paper, the stability aspects of the AHWR are addressed, its three-time-scale properties are brought out and a state feedback regulator for spatial control is designed exploiting the three-time-scale property of the AHWR. © 2011 IEEE.