Individual differences in metacognitive accuracy are generally thought to reflect differences in metacognitive ability. If so, memory monitoring performance should be consistent across different metacognitive tasks and show high test-retest reliability. Two experiments examined these possibilities, using four common metacognitive tasks: ease of learning judgments, feeling of knowing judgments, judgments of learning, and text comprehension monitoring. Alternate-forms correlations were computed for metacognitive accuracy (with a l-week interval between tests). Although individual differences in memory and confidence were stable across both sessions and tasks, differences in metacognitive accuracy were not. These results pose considerable practical and theoretical challenges for metacognitive researchers.The nature of memory monitoring processes has been studied with increasing interest. Most investigations focus on metacognitive accuracy, the relationship between metacognition and future memory performance. There is now a broad base ofempirical observations confirming the accuracy of such judgments, with a wide variety of tasks (see Metcalfe & Shimamura, 1994;Nelson, 1992;Reder, 1996). Investigators in the field have begun to move toward a synthesis ofthese findings; indeed, several theoretical accounts of metacognitive performance have been proposed