As the Alzheimers disease field moves to studies and intervention trials in the preclinical phase and early prodromal period, it will be necessary to measure everyday function in an increasingly more sensitive and sophisticated way to capture more subtle impairments. One approach to increasing sensitivity in functional measures is to use performance based instruments, such as the UCSD Performance-based Skills Assessment (UPSA), in Mild Cognitive Impairment (MCI) or mild Alzheimers disease (AD) research. In this test patients are observed and their response scored as they actually perform proxies for real world tasks and it contrasts with more typical informant based measures. In a preliminary study we compared patients with MCI, patients with mild AD who by diagnosis have functional impairments, and healthy age matched controls on the UPSA, as well as measures of cognition (e.g., episodic memory, semantic memory, executive function, speed). We found that patients with MCI had compromises in everyday functional competence and that the UPSA was strikingly sensitive to these (Goldberg et al, 2010). Additionally the UPSA outperformed an informant based measure on a variety of criteria. However, that study was not longitudinal. Therefore, it is important that we obtain data on the longitudinal characteristics of the UPSA in these populations, including the severity of decline in this measure over time, the relationship of decline to cognitive changes in order to determine the validity of the UPSA, and its technical psychometric characteristics (e.g., test-retest reliability).