Determining total energy expenditure (EE) in children under free-living conditions has become of increasingly clinical interest. The aim of this study was to compare three different methods to assess EE triaxial accelerometry (TriTrac-R3D; Professional Products, Division of Reining International, Madison, WI, USA), activity diary and heart-rate (HR) monitoring combined with indirect calorimetry (IC). Twenty non-obese children and adolescents, aged 5.5 to 16.0 years, participated in this study. Results from the three methods were collected simultaneously under free-living conditions during the same 24 h schoolday period. Neither activity diary (5904 (SD 1756) KJ) NOR THE TRITRAC-R3D (6389 (sd 979) kJ) showed statistical differences in 24 h total EE compared with HR monitoring (5965 (sd 1911) kJ). When considering different physical activity (PA) periods, compared with HR monitoring, activity diary underestimates total EE during sedentary periods (P<0·001) and overestimates total EE and PA-EE during PA periods (P<0·001) because of the high energy cost equivalence of activity levels. The TriTrac-R3D, compared with HR monitoring, shows good agreement for assessing PA-EE during PA periods (mean difference +0·25 (sd 1·9) kJ/min; 95 % CI for the bias -0·08, 0·58), but underestimates PA-EE and it does not show good precision during sedentary periods (-0·87 (sd 1·4) kJ/min, P<0·001). Correlation between the vector magnitude generated by the TriTrac-R3D accelerometer and EE of activities derived from HR monitoring is high. When compared with the HR method, the TriTrac-R3D and activity diary are not systematically accurate and must be carefully used for the assessment of children's EE depending on the purpose of each study.