Ruchkin, Johnson, Mahaffey, and Sutton (1988) presented evidence for a frontal positive/posterior negative late slow wave (SW) which they found to be functionally related to conceptual load, i.e., the difficulty of mental calculation problems increased both the positive and negative parts of it. In the present study we replicated the paradigm of Ruchkin et al. with some modifications, and we also found that this late SW pattern is actually due to a superimposition of two slow potentials. Our results suggest that one potential (positive at frontopolar scalp) is related to the mental operation of division. However, the other potential (negative over posterior scalp) is not related to the computational task itself but to the expectation of stimuli that follow the task. In addition, we found that memorizing a digit seems to be associated with a positive slow wave over posterior scalp. Altogether, our data suggest that load imposed on working memory is associated with positive slow waves which show a task specific topography-mental division is associated with a pSW at F(PZ, remembering with a pSW at P(Z)/O(Z). On the other hand, the state of stimulus and task anticipation is associated with negative slow waves. The latter reach their amplitude maximum over posterior scalp, if visually presented information is anticipated. Our study demonstrates how functionally distinct slow waves can be disentangled by a systematic manipulation of events which either precede or follow the slow wave activity. Moreover, it shows that recording epochs must be of considerable length, if the functional significance of slow waves is the objective of research.