Predicting gestational diabetes by clinical scoring is meant to improve the efficiency of screening. The aim of this study was to determine the usefulness of clinical scoring in a Malaysian population. Using a retrospective cohort of 1997 women, a predictive model was developed and validated on a separate set of 1,000 patients. The efficiency of screening with risk scores was then compared with universal screening. The model derived was well calibrated (p=0.80) with an AUC 0.74 (95 % CI 0.71–0.76). Different combinations of thresholds for the risk score produced a screening reduction between 26 % and 29 % (Sensitivity- 86 % compared to 97 % from universal screening) in Strategy 1. Strategy 2 had no screening reduction but had a higher sensitivity of 95 %. The performance of the risk score was moderate and the screening reduction minimal. Therefore the usefulness of clinical scoring in our population is of limited value.