Background: Biomarkers allow for the early identification and treatment of Alzheimer's Disease (AD) before the onset of symptoms, but may result in longer treatment and associated side effects. Objectives: To investigate heterogeneity of expressed preferences for treatment benefits and risks for delaying the onset of AD symptoms. Methods: Older US adults (n=1004) completed a web-enabled discrete-choice survey where they were to suppose that they would develop AD in the future without medication based on a screening test. Choice tasks presented the option of no treatment vs. a hypothetical AD treatment that would extend time w/ normal memory combined with adverse effects. Choice data were analyzed using scale-adjusted latent-class analysis (LCA). Results: Most participants (66%) were willing to accept relatively high risks of treatment-related stoke and death in the first year of treatment (mean 8-16%) in exchange for 1-2 more years of normal memory. LCA revealed 3 distinct preference classes where membership was correlated with patient characteristics. Conclusions: LCA results are useful to detect heterogeneity of preferences for early disease interception following screening.