Background.- The use of unsuitable statistical techniques to compare statistics calculated on different groups may often result in erroneous conclusions. As it is widely known, while working with small samples (n1 and /or n2< 30) or with samples having non-homogeneous variances, it is much safer to use non-parametric techniques. In this study aiming at comparing the means of two independent samples, Mann-Whitney U test, Smith Satterthwaite and Cochran Cox methods were used. Levels of β carotene, vitamin A and E in two independent groups consisting respectively of 23 patients with breast cancer and of 10 healthy people were compared using the above – mentioned tests. While no significant difference could be found between the levels of β carotene and of vitamin E, the difference betweeen the mean levels of vitamin A was statistically significant.