A crucial aspect of diagnostic reasoning is the integration of sequentially incoming information into a consistent mental representation. While research stresses the importance of working memory in such a task, it is not clear how the information represented in working memory can guide the retrieval of associated information from long-term memory. Factors that might influence this retrieval are the amount of information currently in the focus of attention (Lovett, Daily & Reder, 2000) and the time since the information first became available (Wang, Johnson & Zhang, 2006). By comparing the results of different ACT-R models to human data from a sequential diagnostic reasoning task, we show that these factors do not necessarily influence the retrieval. Our findings rather suggest that in a task where information has to be actively maintained in working memory, each piece of this information has the same potential to activate associated knowledge from long-term memory, independent from the amount of information and the time since it entered working memory.