The study examined the differences in the frequency and quality of student generated questions while using a hypermedia environment. We also examined what specific self-regulatory processes were present prior and following student generated questions. The experiment involved 20 college students from a larger study and categorized them into two separated groups. One group, which was called low shifters, contained participants who had mental model shifts from one to five out of a possible twelve, regardless of whether they started with low or high mental models. The second group, which was called high shifters, contained participants who had mental model shifts above five out of a possible twelve, regardless of whether they started with low or high mental models. Analysis revealed that there were no significant differences in the raw number of questions asked between the participants in the low shifters group and those in the high shifters group. Additional analysis also found that there were no significant differences in the amount of “deep” or “shallow” questions asked by participants in the two groups. In addition, we discovered that there was a significant difference in the specific self-regulatory processes that were used by the high shifters following the generation of a question. More specifically, the role of metacognitive processes is key to understanding the nature of question-generation during tutoring. We propose several implications for the design of intelligent learning environments (ILEs).