In this paper, we discuss the accuracy of several core loss models as compared to experimental data obtained from measurements on commercial hydropower generators. Our aim in this paper is twofold. First, we emphasize the problem of total core loss prediction by comparing the core loss figures obtained by different loss prediction schemes with measured no-load losses. Second, we examine the predicted net effect of bidirectional flux, i.e., the rotational loss. We show that the average degree of rotation in the stator core, and hence the rotational loss, is closely related to the stator slot geometry.