Objectives: In this study, we sought to determine the accuracy of energy expenditure (EE) esti- mation for the Fitbit Charge HR 2 (Fitbit) and the Apple Watch. Design: An observational study. Methods: Thirty young adults (15 men and 15 women, aged 23.5 ± 2.96 years) completed the Bruce treadmill protocol. We measured gross EE by a PARVO metabolic cart (MetCart) and concurrently estimated by the Fitbit and Apple Watch. We calculated concordance correlation coefficients (CCC, rc) and relative error rates to indicate the difference between each device and the MetCart system. Results: For the Apple Watch and Fitbit, the relative error rate was 24.3%, 20.1% for the pooled sample, 18.6%, 24.2% for men, and 29.9%, 16.7% for women, respectively. The Apple Watch overestimated EE for women and underestimated EE for men; the Fitbit underestimated EE for both. Moderate CCCs between estimated EEs and MetCart measured EEs were found for both Apple Watch (rc =0.65, 0.43, and 0.39 overall, men and women, respectively) and Fitbit (rc =0.53, 0.39, and 0.21 overall, men and women, respectively). Conclusion: Neither device showed accurate results compared with EE measured by a MetCart. Users should consider these results when designing programs or personal training plans where physical activity EE is a key outcome assessed with a wearable device.