UTA-GMS is a new method for assessment of strong or weak outranking relation in a problem of multi-criteria ranking, proposed by the authors. The ranking concerns a finite but relatively large set of alternatives A. We assume indirect preference information supplied by the decision maker (DM) in form of a complete preorder on a subset of reference alternatives R, called reference preorder. The preference model build from this information is an additive value function. The technique of passing from reference preorder to compatible additive value functions is called ordinal regression and it is well known from the UTA method proposed by Jacquet-Lagreze and Siskos in 1982. Unlike in the UTA method, we take into account all compatible value functions (instead of one or several most characteristic) at the stage of ranking the whole set A of alternatives. Moreover, we do not impose the additive value function to have piecewise-linear components but we accept any additive form. The resulting relations in A are twofold: strong outranking (if alternative x has greater value than y for all compatible value functions) and weak outranking (if alternative x has greater value than y for at least one compatible value function). Strong outranking is a partial preorder and weak outranking is a complete preorder in A. The strong outranking is of particular interest for the DM – it corresponds to dominance relation when the set of reference alternatives is empty, and to a complete preorder relation when the reference ranking is compatible with a single value function only. This approach has several interesting extensions useful for practical applications. The method has been implemented for a PC and will be presented together with an example of application.