A great variety of multi-criteria decision aiding (MCDA) methods has already been developed but few papers have dealt with mixed data (qualitative and quantitative). MCDA techniques accepting different types of evaluations (such as deterministic, stochastic and/or fuzzy ones) are rather rare and not very well known, even though this issue is crucial from a practical point of view, since mixed evaluations occur very frequently in appraising and selecting projects and organizations, as well as in risk management modelling, among other fields. This paper presents a new discrete MCDA tool developed for mixed performances of alternatives called BIPOLAR MIX. It is based on the classical BIPOLAR method proposed by Konarzewska-Gubała (1989), and on its modification, namely the BIPOLAR method with stochastic dominance (SD) rules, proposed by Górecka (2009). A numerical example at the end of the paper illustrates the problem of ordering projects applying for co-financing from the European Union (EU).