The contribution presents the results of confirmatory factor analysis of the Rosenberg Self-Esteem Scale (RSES) among Slovak subjects. The sample consisted of 920 high school and university students (365 males, 555 females). Using LISREL software, eight different models of factor structure were estimated. The models were formulated on the basis of the previous results, and they included a one-factor model, a one-factor model with correlated uniquenesses of negatively worded items, an one-factor model with correlated uniquenesses of positively worded items, four two-factor models (positive and negative self-esteem, self-liking and self-competence, general and transient self-esteem, self-derogation and defense of self), and a three-factor model (social comparison, positive and negative self-esteem factors). The best fit indices were found for the model one-factor with correlated uniquenesses of negatively worded items representing unidimensional construct contaminated by effect associated with negatively worded items. Satisfactory fit indices were also found for the one-factor model with correlated uniquenesses of the positively worded items, for the two-factor solution with positive and negative self-esteem and for the three factor solution. Correlational analysis revealed that factors based on these models differ in their correlations with Big Five personality traits as measured by NEO-FFI. This suggests that Rosenberg Self-Esteem Scale can be used alternatively as one, two or three dimensional measure.