OBJECTIVES:To explore the atrophy rate of entorhinal cortex (ERC) in AD and normal aging and assess the value of rate measurement of ERC atrophy for classifying subjects with AD from cognitively normal (CN) control subjects. METHODS:Twenty-one AD patients and 23 CN subjects had MRI scans and clinical evaluations twice within 1.8 +/- 0.6 years. ERC volumes were manually measured on volumetric T1-weighted MR images. RESULTS:Patients with AD had a greater annual percentage volume change of ERC than CN subjects on both sides (left: 6.8 +/- 4.3%/year for AD vs 1.4 +/- 2.5%/year for CN [F(1,42) = 25.6, p < 0.001]; right: 6.3 +/- 3.3%/year for AD vs 1.4 +/- 2.3%/year for CN [F(1,42) = 25.6, p < 0.001]). Furthermore, increased ERC atrophy rate was correlated (r = -0.56, p = 0.01) with decreased memory performance in AD. CN subjects had on average annual ERC atrophy rates greater than zero (p < 0.01). Baseline volume of ERC predicted atrophy rate of ERC (left: r = -0.53, p < 0.01; right: r = -0.42, p < 0.05) in CN subjects but not in AD subjects. Using ERC baseline volumes alone resulted in 77% overall correct classification (p < 0.01) between AD and CN subjects, with 76% sensitivity and 78% specificity and an area under receiver operator characteristic (ROC) curve of 0.83. Adding annual atrophy rate of ERC to the model accounted for most of the variance (p < 0.01), diminishing contributions from baseline volume and yielding 82% overall classification, with 76% sensitivity and 86% specificity and an area under the ROC curve of 0.93. CONCLUSION:ERC volume loss over time may be a better indicator for AD than cross-sectional measurements.