Recognition memory tests are useful for understanding Alzheimer’s disease (AD). In the clinical context, it is important to model performance at both the group level (e.g., for the characterization of clinical subpopulations) and individual level (e.g., for the diagnosis of a patient). Using a clinical data set from AD patients, we show how a signal detection theory model that assumes hierarchical individual differences in discriminability and response bias adequately describes these data at both the group and individual levels, and also present preliminary descriptive and predictive analyses of the data at both levels.