This investigation examines the effects of keyword tasks (Immediate vs. Delayed) on metacognitive monitoring, study regulation, and recall in multi-step learning tasks, which require learning information from expository texts. The titles of the expository texts were biased towards information that was either stated close to the title (Related/Close), distant from the title (Related/Distant), or unrelated to the title (Unrelated). Based on the Cue-Utilization Framework, we hypothesized that learners’ metacognitive monitoring and study regulation would be informed by mnemonic cues derived from text-titles and keyword tasks. Two hundred and thirteen American undergraduate students studied six expository texts, generated keywords, provided judgments of learning, and wrote about what they recalled before and after a self-regulated restudy trial. In line with our main hypothesis, the results revealed that learners who generated keywords immediately overestimated their current state of learning to a greater extent than learners who generated keywords with a delay. Contrary to our expectations, the greater monitoring accuracy observed in the delayed keyword group did not result in more effective restudy behavior. Learners in both keyword groups were able to improve their recall performance from their first to their second set of recall tasks, but interestingly, only learners in the immediate keyword group utilized the restudy trial to close knowledge gaps between information, which was stated close to versus distant from the title.