OBJECTIVE This study aimed to measure the metabolic cost (METs) of walking activities in older adults, to examine the relationship between accelerometer output and METs across walking activities, and to compare measured MET values in older adults with the MET values in the compendium. METHODS Twenty older adults (mean age = 75, range = 60-90 yr) completed eight walking activities (five treadmill based, three free living) for 6 min each. Oxygen consumption (V˙O2) and resting metabolic rate (RMR) were measured using a portable metabolic system, and motion was recorded using a waist-mounted ActiGraph accelerometer (GT3X; ActiGraph, Pensicola, FL). Energy expenditure across activities was defined as kilocalories per minute and measured as METs (V˙O2 / RMR) and standard METs (V˙O2 / 3.5 mL·kg-1·min-1). Mixed modeling was used to assess differences in counts per minute and kilocalories per minute by weight status, sex, comorbidity status, and functional status. Linear regression analysis was applied to develop a prediction equation for kilocalories per minute. Energy costs of walking were subsequently compared with METs in the compendium of physical activities. RESULTS Average measured RMR was 2.6 mL·kg-1·min-1, 31.6% less than the standard RMR of 3.5 mL·kg-1·min-1. On average, standard METs were 71% lower than the measured METs across all walking activities. Measured MET levels differed from previously reported values in the literature and values listed in the compendium, resulting in misclassification of activity intensities for 60% of the walking conditions. Average counts for the walking activities ranged from 809 (treadmill = 1.5 mph) to 4593 counts per minute (treadmill = 3.5 mph). Previous regression equations consistently overestimate all activities compared with the measured energy cost in this sample of older adults. CONCLUSIONS This study identifies the need for equations and cut points specific to older adults.