In this paper, we consider the analysis of systems with parametric uncertainties. Due to the intrinsic complexity of this problem, the proposed efficient analysis methods are based on sufficient thus possibly conservative conditions. A popular one, in the mu and in the quadratic constraint approaches, involves the use of D and G scalings. In this paper, we reveal how to reduce the conservatism of the D-G scaling analysis while considering more general structures of parametric uncertainties and keeping numerical efficiency. Numerical examples emphasize the benefit of the proposed extension