The aim of this study was to examine the characteristics of diurnal cortisol rhythm in childhood obesity and its relationships with anthropometry, pubertal stage and physical activity. Thirty-five children with obesity (median age: 11.80[interquartile range 10.30, 13.30] and median BMI z-score: 3.21[interquartile range 2.69, 3.71]) and 22 children with normal weight (median age: 10.85[interquartile range 8.98, 12.13] and median BMI z-score: − 0.27[interquartile range − 0.88, 0.35]) were recruited. Saliva samples were collected at 08:00, 16:00 and 23:00 h. Cortisol concentrations at 3 time points, corresponding areas under the curve (AUCs) and diurnal cortisol slope (DCS) were compared between the two groups. Anthropometric measures and pubertal stage were evaluated, and behavioural information was obtained via questionnaires. Children with obesity displayed significantly lower cortisol08:00 (median [interquartile range]: 5.79[3.42,7.73] vs. 8.44[5.56,9.59] nmol/L, P = 0.030) and higher cortisol23:00 (median [interquartile range]: 1.10[0.48,1.46] vs. 0.40[0.21,0.61] nmol/L, P < 0.001) with a flatter DCS (median [interquartile range]: − 0.29[− 0.49, 0.14] vs. -0.52[− 0.63, 0.34] nmol/L/h, P = 0.006) than their normal weight counterparts. The AUC increased with pubertal development (AUC08:00–16:00:P = 0.008; AUC08:00–23:00: P = 0.005). Furthermore, cortisol08:00 was inversely associated with BMI z-score (β = − 0.247, P = 0.036) and waist-to-height ratio (WHtR) (β = − 0.295, P = 0.027). Cortisol23:00 was positively associated with BMI z-score (β = 0.490, P<0.001), WHtR (β = 0.485, P<0.001) and fat mass percentage (FM%) (β = 0.464, P<0.001). Absolute values of DCS were inversely associated with BMI z-score (β = − 0.350, P = 0.009), WHtR (β = − 0.384, P = 0.004) and FM% (β = − 0.322, P = 0.019). In multivariate analyses adjusted for pubertal stage and BMI z-score, Cortisol08:00, AUC08:00–16:00 and absolute values of DCS were inversely associated with the relative time spent in moderate to vigorous intensity physical activity (P < 0.05). AUC16:00–23:00 was positively associated with relative non-screen sedentary time and negatively associated with sleep (P < 0.05). The disorder of diurnal salivary cortisol rhythm is associated with childhood obesity, which is also influenced by puberty development and physical activity. Thus, stabilizing circadian cortisol rhythms may be an important approach for childhood obesity.