This paper presents a new approach to the analysis of the stability robustness of dynamic systems in state-space models. By continuously perturbing the state matrix, instead of solving the Lyapunov equation, an elegant stability robustness fundamental is derived. Subsequently, the allowable norm bound of the error matrix can be obtained under weakly structured perturbations, and the magnitude bound on individual elements of the error matrix can be obtained under highly structured perturbations. The merits of the theorems and corollary developed are demonstrated by two examples where the results achieved are much better than those already published in the literature. The concept that the perturbed state matrix would depend on the operating frequencies is also introduced.