Working memory (WM), our central system for temporarily binding task-relevant information to the current context for ongoing processing, is limited in capacity. To overcome such limitations, we frequently leverage prior knowledge from long-term memory (LTM), allowing us to integrate and represent information more efficiently. Yet, it is not well understood how prior knowledge affects the representation of information in working memory. Here, we assessed how prior learning affects the load of a visual WM representation by using multivariate load classification from EEG. Participants (N=30) learned a 6-color visual array to criterion, and then completed a WM task including both new and pre-learned arrays. Crucially, new arrays differed in set size (0, 1, 2 or 6), which we used to train a classifier to identify WM load from the multivariate EEG signal. After establishing strong classification accuracy (~55%; chance = 25%), we asked the classifier to predict the load elicited by pre-learned arrays. We find evidence that the availability of LTM for pre-learned arrays reduced WM load, as the classifier predicts a load of 1 or 2, instead of 6 – the actual set size of pre-learned arrays. However, further exploration revealed that LTM representations were still dissociable from pure WM representations. Our results suggest that the availability of prior knowledge can reduce WM load by reducing the number of bindings that have to be maintained but leads to qualitative different representations as indicated by distinct changes in multivariate neural signals.