This paper deals with the effects that the popular normalisation procedures: Simple (SN), Linear (LN). and Vector (VN) could have on the results of the Multiple Attribute Decision Making (MADM) methods (Simple Additive Weighting-SAW, TOPSIS and ELECTRE). It shows that the deformations of empirical data, caused by the usage of normalisations, could affect the final choices. lf the MADM methods are based on normalised ratings of the SN or VN type, then their results could depend on: 1) the measurement units used for quantitative attributes that are measurable by interval scales, and 2) the type of Likert scale used for measuring qualitative attributes. It also shows that the MADM methods violate certain conditions of consistent choice and that this violation could be attributed to SN, LN and VN normalisation procedures.