The purpose of this study was to evaluate the potential for using accelerometer-determined ambulatory activity indicators (steps per day and cadence) to predict total energy expenditure (TEE) and physical activity energy expenditure (PAEE) derived from doubly labeled water (DLW). Twenty men and 34 women (20-36 years of age) provided complete anthropometric, accelerometer, resting metabolic rate (RMR), and DLW data. TEE and PAEE were determined for the same week that accelerometers were worn during waking hours. Accelerometer data included mean steps per day, peak 30-min cadence (average steps per minute for the highest 30 min of the day), and time spent in each incremental cadence band: 0 (nonmovement), 1-19 (incidental movement), 20-39 (sporadic movement), 40-59 (purposeful steps), 60-79 (slow walking), 80-99 (medium walking), 100-119 (brisk walking), and 120+ steps·min(-1) (indicative of all faster ambulatory activities). Regression analyses were employed to develop sex-specific equations for predicting TEE and PAEE. The final model predicting TEE included body weight, steps per day, and time in incremental cadence bands and explained 79% (men) and 65% (women) of the variability. The final model predicting PAEE included peak 30-min cadence, steps per day, and time in cadence bands and explained 76% (men) and 46% (women) of the variability. Time in cadence bands alone explained 39%-73% of the variability in TEE and 30%-63% of the variability in PAEE. Prediction models were stronger for men than for women.