Working memory performance is often assumed to benefit from different maintenance control strategies such as rehearsal, refreshing, elaboration, and grouping. In studies assessing strategy self-reports, some strategies were indeed associated with better recall. Nevertheless, experimental studies assessing the effect of instructing maintenance strategies compared to a no-instruction baseline lend no evidence for the effectiveness of these strategies for working memory. Explanations for this contradiction could be that instruction implementation engenders dual-task costs, or that strategy instructions reduce adaptive strategy switching. Across two experiments, we investigated the frequency and variability of strategy use with trial-wise self-reports in serial recall of word lists. Further, we examined potential instruction costs by comparing performance in trials with self-reported vs. instructed use of the same strategies. Self-reported strategy use varied from trial to trial, with elaboration and rehearsal being the most frequent. Self-reported elaboration was correlated with better performance than reading and rehearsal. For the most prevalent strategies – elaboration and rehearsal – there were no costs of instructed strategy-implementation. Our results speak against dual-task costs, and for an advantage of adaptively choosing one’s own strategy from trial to trial.