This study was conducted to determine the accuracy of learners' comprehension monitoring during computer-based instruction and to assess the relationship between enroute monitoring and different levels of learning. Participants were 50 university undergraduate students enrolled in an introductory educational psychology class. All students received the same treatment--four sections of computer-based instruction designed to require learners to sort among numerous facts and concepts. Following each section, students were asked to rate their level of understanding of both the factual and the inferential materials. They then answered eight short-answer embedded questions--four factual and four inferential. Factual questions were divided into three levels of specificity and detail; inferential questions were divided into intraframe and interframes inferences. A posttest of 32 items was also administered; half of the questions were repeated from the lesson and half were new items. Regression procedures were conducted to determine the predictive value of enroute self-assessments for corresponding factual and inferential learning. Enroute ratings were used to predict student performance on embedded and posttest questions, as well as to examine the relationship among ratings for facts or inferences. Little or no correlation was found between the ratings of understanding and subsequent performance on either the embedded questions or the posttest, suggesting that learners may not be good judges of their enroute comprehension. A list of references and nine data tables complete the study. (JB)