Prior research has shown that people are likely to skim information presented digitally with the resultant deleterious effect on accurate mental models of the text. Teaching monitoring strategies and presenting text with adjunct questions are effective strategies for improving the mental models of readers of scientific text, but the two strategies have not been directly compared for studying text read from a computer screen. In the present study, participants were individually tested in two experiments in which they studied computer-based text on the heart and circulatory system while either reading the text with no interruption, answering adjunct inference questions, answering adjunct factual questions, or answering prompts for global or specific monitoring. Reading time was untimed but looking back was not available. Prompting for global monitoring was most effective for improvement of mental models as measured by scores in diagram drawing, concept map construction, factual memory, and generating inferences. Further, global monitoring did not require more time-on-task than the other methods. (Contains 8 tables and 1 figure.)