Objectives and study In the development of Celiac Disease (CD) both genetic and environmental factors play a crucial role. The Human Leukocyte Antigen (HLA)-DQ2 and HLA-DQ8 loci are strongly related to the disease and are necessary but not sufficient for the development of CD. Therefore, increasing interest lies in examining the mechanisms of CD onset from the early beginning. Differences in serum and urine metabolic profiles between healthy individuals and CD patients have been reported previously. We aimed to investigate if the metabolic pathways were already altered in young, 4 month old infants, preceding the CD diagnosis. Methods Serum samples were available for 230 four month old infants of the PreventCD project, a multicenter, randomized, double-blind, dietary intervention study. All children were positive for HLA-DQ2 and/or HLA-DQ8 and had at least one first-degree relative diagnosed with CD. Amino acids were quantified after derivatization with liquid chromatography – tandem mass spectrometry (MS/MS) and polar lipid concentrations (acylcarnitines, lysophosphatidylcholines, phosphatidylcholines, and sphingomyelins) were determined with direct infusion MS/MS. We investigated the association of the metabolic profile with (1) the development of CD up to the age of 8 years (yes/no), (2) with HLA-risk groups, (3) with the age at CD diagnosis, using linear mixed models and cox proportional hazards models. Gender, intervention group, and age at blood withdrawal were included as potential confounder. Results By the end of 2014, thirty-three out of the 230 children (14%) were diagnosed with CD according to the ESPGHAN criteria. Median age at diagnosis was 3.4 years (IQR, 2.4–5.2). Testing each metabolite for a difference in the mean between healthy and CD children, we (1) could not identify a discriminant analyte or a pattern pointing towards an altered metabolism (Bonferroni corrected P > 0.05 for all). Metabolite concentrations (2) did not differ across the HLA-risk groups. When investigating the age at diagnosis using (3) survival models, we found no evidence for an association between the metabolic profile and the risk of a later CD diagnosis. Conclusion The metabolic profile at 4 months of age was not predictive for the development of CD up to the age of 8 years. Our results suggest that metabolic pathways reflected in serum are affected only later in life and that the HLA-genotype does not influence the serum metabolic profile in young infants before introduction of solid food.