The pathophysiology of autistic spectrum disorder (ASD) is not fully understood and there are no diagnostic or predictive biomarkers. Glycosylation modified as many as 70% of all human proteins can sensitively reflect various pathological changes. However, little is known about the alterations of glycosylation and glycoproteins in ASD. In this study, serum glycopattern and the maackia amurensis lectin-II binding glycoproteins (MBGs) in 65 children with ASD and 65 age-matched typically developing (TD) children were compared by using lectin microarrays and lectin-magnetic particle conjugate-assisted LC-MS/MS analyses. Expression of Siaα2-3 Gal/GalNAc was significantly increased in pooled (fold change = 3.33, p < 0.001) and individual (p = 0.009) serum samples from ASD versus TD children. A total of 194 and 217 MGBs were identified from TD and ASD sera respectively, of which 74 proteins were specially identified or up-regulated in ASD. Bioinformatic analysis revealed abnormal complement cascade and aberrant regulation of response-to-stimulus that might be novel makers or markers for ASD. Moreover, increase of APOD α2-3 sialoglycosylation could sensitively and specifically distinguish ASD samples from TD samples (AUC is 0.88). In conclusion, alteration of MBGs expression and their sialoglycosylation may serve as potential biomarkers for diagnosis of ASD, and provide useful information for investigations into the pathogenesis of ASD.