The prevalence of fetal macrosomia is steadily increasing worldwide, reaching up to 20%. Fetal macrosomia complicates pregnancy and delivery. Current prediction strategies are inaccurate, and most patients with fetal macrosomia go into labor with an "unknown status." The aim of this study was to develop a system for predicting fetal macrosomia based on the lipid profile of the pregnant woman's blood serum. 110 patients were included in the study, 30 patients had gestational diabetes mellitus (GDM), 80 – had not. During the observation, blood samples were collected at three time points: in the first trimester (11-13 weeks of pregnancy), in the second trimester (24-26 weeks), and in the third trimester (30-32 weeks). Lipids were detected by flow injection analysis with mass spectrometry. Lipid profiles of pregnant women were discriminated by orthogonal projection on latent structure discriminant analysis (OPLS-DA) in all three trimesters. The developed OPLS-DA models allow predicting the occurrence of fetal macrosomia during pregnancy. Three sets of models were developed: models independent of GDM status with sensitivity of 0.85, and specificity of 0.91; models for patients with positive GDM status with sensitivity of 0.91, and specificity of 0.96; models for patients with negative GDM status with sensitivity of 0.93, and specificity of 0.92. Phosphatidylcholines and sphingomyelins were the most important discriminative features. These lipid groups probably play an important role in the pathogenesis of fetal macrosomia and may serve as laboratory markers of this pregnancy complication.