Summary: In most of the Goal Programming (GP) applications reported in the literature the election of the GP variant is made in a mechanistic way by the analyst. However, each variant implies a different preference structure and, thus, it should be chosen by the decision-maker. Moreover, the final solution of the problem depends critically on the variant chosen. In this paper, it is assumed that the use of a combination of GP variants may reflect in many cases more precisely the actual preferences of the decision-maker. In this way, a “sensitivity analysis” of the solution obtained as well as of the own model structure can be very helpful to let the decision-maker clarify his/her knowledge about the reality analysed. This type of problem requires a more general approach that leads to a GP formulation of the initial GP model. This type of meta-GP approach is presented in this paper.