The stability and accuracy of memory perceptions in 2 longitudinal samples was examined. Sample 1 consisted of 231 adults (22-78 years) tested twice over 2 years. Sample 2 consisted of 234 adults (55-86 years) tested 3 times over 6 years. Measures of perceived and actual memory change were obtained. A primary focus was whether perceptions of memory change stem from application of an implicit theory about aging and memory or from accurate monitoring of actual changes in performance. Individual differences in metamemory were highly stable over time. Results suggested at least some accurate monitoring of memory in Sample 2, in which actual change was greatest. However the overall pattern of results is largely consistent with predictions derived from an implicit theory hypothesis.