In this work, we describe a multivariate data analysis approach for data exploration and classification of the complex and large data sets generated to study the alteration of human transferrin (Tf) N-glycopeptides in patients with congenital disorders of glycosylation (CDG). Tf from healthy individuals and two types of CDG patients (CDG-I and CDG-II) is purified by immunoextraction from serum samples before trypsin digestion and separation by capillary liquid chromatography mass spectrometry (CapLC-MS). Following a targeted data analysis approach, partial least squares discriminant analysis (PLS-DA) is applied to the relative abundance of Tf glycopeptide glycoforms obtained after integration of the extracted ion chromatograms of the different samples. The performance of PLS-DA for classification of the different samples and for providing a novel insight into Tf glycopeptide glycoforms alteration in CDGs is demonstrated. Only six out of fourteen of the detected glycoforms are enough for an accurate classification. This small glycoform set may be considered a sensitive and specific novel biomarker panel for CDGs.