A central role for neuropsychological evaluations is the measurement of change in cognitive functioning over time. However, change scores obtained from repeated neuropsychological assessments may be affected by normal variability because of measurement error and practice effects caused by repeated measurements. The current study uses reliable change estimates to establish normative rates of change on the Dementia Rating Scale from baseline to first follow-up testing among 1080 cognitively normal adults aged 65 and older. Results showed that a 6-point decline by European Americans or a 9-point decline by African American adults within a 9-15 month test-retest interval represents reliable change. Within a 16-24-month test-retest interval, a 7-point decline among European Americans or an 8-point decline among African American adults represents reliable change. In addition, preliminary cross-validation was performed in a clinical comparison sample of another 22 older adults. The findings are discussed in the context of potential clinical and research applications.