As research increasingly focuses on preclinical stages of Alzheimer’s disease (AD), instruments must be retooled to identify early cognitive markers of AD. A supplemental delayed recall subtest for the Alzheimer’s Disease Assessment Scale-cognitive (ADAS-cog; Mohs, Rosen, & Davis, 1983; Rosen, Mohs, & Davis, 1984) is commonly implemented, but it is not known precisely where along the spectrum of cognitive dysfunction this subtest yields incremental information beyond what is gained from the standard ADAS-cog, or whether it can improve prediction of functional outcomes. An item response theory approach can analyze this in a psychometrically rigorous way. Seven hundred eighty-eight patients with AD or amnestic complaints or impairment completed a battery including the ADAS-cog and 2 activities of daily living measures. The delayed recall subtest slightly improved the ADAS-cog’s measurement precision in the mild range of cognitive dysfunction and increased prediction of instrumental activities of daily living for individuals with subjective memory impairment.