Recent evidence highlights the role of adiponectin in the pathophysiology of autism spectrum disorder (ASD), yielding conflicting results. The aims of this study were (1) To assess the adiponectin levels of children with ASD and typical developing (TP); (2) To investigate the relationship between adiponectin levels and symptom severity of children with ASD. This is a single-center cross-sectional study from China. From December 2017 to November 2019, first-diagnosis and drug-naïve children with ASD were included. Same TP children who were matched with clinical groups by gender and age were included as the control group. The enzyme-linked immunosorbent assay (ELISA) kit was used to determine serum concentrations of adiponectin. We recorded 176 children (88 were ASD and 88 were TP children) and 77.3% (n = 136) were boys and the mean age was 4.3 years (standard deviations [S.D.]: 1.2). The mean (S.D.) levels of adiponectin were 9.01(2.19) and 11.55(2.32) μg/ml for those with ASD and TP subjects. The difference between those two groups was significant (t = 7.169, p < 0.001). There was a negative correlation between serum levels of adiponectin and Childhood Autism Rating Scale [CARS] score (r = −0.498, p < 0.001). At admission, 39 ASD (54.5%) had a minor autism (CARS<37). In these children, the mean (S.D.) adiponectin level was higher than that observed in children with moderate-to-severe clinical severity (10.09[2.32] vs.8.15[1.64] μg/ml, P < 0.001). This study shows that serum adiponectin. Levels are decreased in ASD when compared with in healthy children. The findings also indicate an inverse association between serum adiponectin levels and severity of symptoms in ASD.