Background: GDM is associated with adverse maternal and fetal complications. By the time GDM is diagnosed, continuous exposure to the hyperglycaemic intrauterine environment can adversely affect the fetus. Hence, early pregnancy prediction of GDM is important. Aim: To systematically evaluate whether composite risk score prediction models can accurately predict GDM in early pregnancy. Method: Systematic review of observational studies involving pregnant women of <20 weeks of gestation was carried out. The search involved various databases, grey literature, and reference lists till August 2022. The primary outcome was the predictive performance of the models in terms of the AUC, for <14 weeks and 14-20 weeks of gestation. Results: Sixty-seven articles for <14 weeks and 22 for 14-20 weeks of gestation were included (initial search - 4542). The sample size ranged from 42 to 1,160,933. The studies were from Canada, USA, UK, Europe, Israel, Iran, China, Taiwan, South Korea, South Africa, Australia, Singapore, and Thailand. For <14 weeks, the AUC ranges were 0.59-0.88 and 0.53-0.95, respectively for models that used only maternal characteristics and for those that included biomarkers. For 14-20 weeks these AUCs were 0.68-0.71 and 0.65-0.92. Age, ethnicity, BMI, family history of diabetes, and prior GDM were the 5 most commonly used risk factors. The addition of systolic BP improved performance in some models. Triglycerides, PAPP-A, and lipocalin-2, combined with maternal characteristics, have the highest predictive performance. AUC varied according to the population studied. Pooled analyses were not done due to high heterogeneity. Conclusion: Accurate GDM risk prediction may be possible if common risk factors are combined with biomarkers. However, more research is needed in populations of high GDM risk. Artificial Intelligence-based risk prediction models that incorporate fetal biometry data may improve accuracy.