The purpose of this study was to determine the validity of a submaximal exercise test, the Step Test Exercise Prescription (STEP), in a broad age range and in individuals in the earliest stages of Alzheimer's disease (AD). Individuals (n = 102) underwent treadmill-based maximal exercise testing and a STEP. The STEP failed to predict peak oxygen consumption (VO2peak), and was a biased estimate of VO2peak (p < .0001). Only 43% of subjects' STEP results were within 3.5 ml · kg-1 · min-1 of VO2peak. When categorized into fitness levels these 2 measures demonstrated moderate agreement (kappa = .59). The validity of the STEP was not supported in our participants, including those with AD. The STEP may not be appropriate in the clinic as a basis for exercise recommendations in these groups, although it may continue to have utility in classifying fitness in research or community health screenings.