The study aimed to assess a self-reported questionnaire about physical activity (PA) from the Health Behaviour in School-aged Children (HBSC) using triaxial accelerometer as a criterion. Elementary school boys (N=292) and girls (N=313) in the fourth grade and higher completed the questionnaire and wore an accelerometer for at least 10 h/day for at least 4 days. The phi coefficients of the chi-square test 2-way tables (active/inactive as measured using the HBSC questionnaire × achievement/non-achievement of the recommended 60-min moderate to vigorous PA [MVPA] as measured using the accelerometer) were 0.25 (P<0.001), 0.17 (P=0.009), and 0.08 (P=0.217) for all children, boys, and girls, respectively. The sensitivity and specificity in boys were 82% (95% confidence interval [CI]=78-86%) and 34% (95%CI=26-41%) respectively, while the those of girls were 52% (95%CI=40-63%), 57% (95%CI=55-60%). The positive predictive value in boys was 70% (95%CI=66-73%) and negative predictive value in girls was 83% (95%CI=79-87%). The independent t-test showed that moderate PA (MPA), vigorous PA (VPA), and MVPA min/day of active children were significantly higher than those of inactive children (range of Cohen’s d=0.38 to 0.71). Area under the curve (AUC) of the VPA (AUC = 0.60–0.73) in girls was significantly higher than that of the MPA (AUC = 0.52–0.65) and MVPA (AUC = 0.54–0.67). Our results supported that the HBSC self-reported questionnaire has acceptable, but limited agreement for assessing achievement/non-achievement of the MVPA recommendation, and could estimate the differences in the MPA, VPA, and MVPA min/day of children.