The widely-used Brief Self-Control Scale (BSCS) is usually applied as a unidimensional measure of trait self-control. However, there is no clear empirical evidence for the scale's unidimensional structure, while different multidimensional conceptualizations of the BSCS have recently been suggested. The authors of those multidimensional models used different BSCS item subsets to specify distinct facets of self-control in order to enhance the representation of the scale's internal structure or to increase the instrument's efficiency in predicting various outcomes. Up until now, little is known about the relative performance of these conceptualizations. In this article, we compare three two-dimensional representations of the BSCS with the unidimensional measure in two samples of university students (N=205) and apprentices in vocational education and training (N=1951). Of the two-dimensional models only the one that separates positively and negatively worded items showed a consistent improvement in model fit in both samples, compared to the unidimensional model. However, in comparison to the unidimensional measure, the two-dimensional measures did not substantially enhance the predictive power concerning outcome variables in either sample. We conclude that the BSCS's total score is a viable option for assessing trait self-control and for studying its relationship with achievement-related outcome variables.