We propose a methodology that can be useful to the hazard expert in building ground-motion logic trees to capture the center and range of ground-motion amplitudes. The methodology can be used to identify a logic-tree structure and weighting scheme that prevents the dominancy of a specific ground-motion model. This strategy can be useful for regional probabilistic seismic hazard since logic-trees biased by a specific ground-motion predictive model (GMPE) may cause disparities in the seismic hazard for regions represented by large number of sites with complex seismic features. The methodology first identifies a suit of candidate ground-motion prediction equations that can cover the center, body and range of estimated ground motions. The GMPE set is then used for establishing alternative logic-trees composed of different weighting schemes to identify the one(s) that would not be biased towards a particular GMPE due to its sensitivity to the weights. The proposed methodology utilizes visual and statistical tools to assess the ground motion distributions over large areas that makes it more practical for regional hazard studies.