This study establishes tri-axial activity count (AC) cut-points for the GT3X+ accelerometer to classify physical activity intensity in overweight and obese adults. Further, we examined the accuracy of established and novel energy expenditure (EE) prediction equations based on AC and other metrics. "Part 1": Twenty overweight or obese adults completed a 30 minute incremental treadmill walking protocol. Heart rate (HR), EE, and AC were measured using the GT3X+ accelerometer. "Part 2": Ten overweight and obese adults conducted a self-paced external walk during which EE, AC, and HR were measured. Established equations (Freedson et al., 1998; Sasaki et al., 2011) overestimated EE by 40% and 31%, respectively (p < 0.01). Novel gender-specific prediction equations provided good estimates of EE during treadmill and outdoor walking (standard error of the estimate = 0.91 and 0.65, respectively). We propose new cut-points and prediction equations to estimate EE using the GT3X+ tri-axial accelerometer in overweight and obese adults.