One of the major challenges in scientific research is to understand past climate and the mechanisms of climate change. Small vertebrates, and especially rodents, are very sensitive to shifts in climate and habitat, and their variations over time in terms of taxa and abundance can be successfully used to reconstruct past environments. The vast array of approaches to palaeoclimatic reconstruction reflects the great effort that has gone into this line of investigation. Recently, the UDA-ODA discrimination technique has been postulated as a more reliable ecologically-based methodology compared to the classical MER method. To provide biogeographical information to be analysed by the UDA-ODA discrimination technique, the distributions of four species (Sorex minutus, Chionomys nivalis, Talpa europaea and Crocidura russula) documented in levels O, N, E and D of the Abric Romaní site were processed. The results reveal a statistical difference between the climatic values for the occupied distribution areas (ODA) and those for the uncertain distribution areas (UDA). This technique was then applied to small-mammal assemblages from the above-mentioned levels of Abric Romaní, to test whether the use of the ODAs of the species improves the precision of the climatic reconstruction compared to the atlas distributions of the species used in MER procedures. Our results suggest an improvement in the discrimination analysis over the previous MER reconstructions when wider distributions for an assemblage are obtained. The coldest values obtained for level O of Abric Romaní seem to reinforce the pollen interpretation of the level as coetaneous with a cold period. For the whole MIS 3 climatic scenario for Neanderthals, a colder and wetter climate is derived from the small-mammal analysis. However, as different methods and analyses have inherent limitations, a standardization of the methods applied to the different levels and sites should be carried out in order to provide comparable results