Subjective memory change (SMC) in adulthood involves the perception that one’s memory has declined from earlier levels of function. SMC has been conjectured to be more accurate than concurrent subjective memory because people use themselves as a standard of comparison. We used data from two longitudinal studies to contrast the accurate-monitoring-of-change hypothesis—actual memory change predicts SMC—against a constructed-judgment hypothesis that rated SMC is a function of rescaling concurrent memory beliefs without accessing actual memory change. It states that actual memory change has no predictive validity for SMC independent of concurrent memory beliefs. Data from both the Berlin Aging Study and the Health and Retirement Study (HRS) showed that older adults’ current memory complaints strongly predicted current SMC, and that there was little relationship of longitudinally measured memory change to SMC, controlling on memory complaints. In the HRS there were reliable latent-growth-curve slope correlations of over .20 for change in episodic memory with both slopes of change in SMC and in memory complaints, yet little relationship of SMC slopes to episodic memory slopes, controlling on memory-complaint slopes. The results falsify the accurate-monitoring-of-change hypothesis regarding the origins of SMC in older adults.