OBJECTIVE Disagreement about the appropriate criteria for the diagnosis of gestational diabetes mellitus (GDM) persists. This study addresses the problem by examining an alternative approach which combines information from all time-points on the glucose tolerance test (GTT) into a single index and expands the GDM spectrum beyond the current binary approach into four categories using data from three geographically and ethnically distinct populations. DESIGN Retrospective observational study design SETTING Data from Wisconsin, USA (723 women) was used in derivation of the criterion and data from Doha, Qatar (1284 women) and Cape Town, South Africa (220 women) for validation. PARTICIPANTS Pregnant women without pre-existing diabetes with a GTT done between 23 and 30 weeks gestation MAIN OUTCOME MEASURE A novel index was derived from the GTT termed the weighted average glucose (wAG). This was categorized into four pre-defined groups (henceforth National Priorities Research Program (NPRP) criterion); i) normal gestational glycemia (NGG), ii) impaired gestational glycemia (IGG), iii) GDM and iv) high risk GDM (hGDM). RESULTS In the Doha cohort, compared to the NGG group, the odds of large for gestational age babies increased 1.33 fold (P=0.432), 2.86 fold (P<0.001) and 3.35 fold (P<0.001) in the IGG, GDM and hGDM groups respectively. The odds of pregnancy induced hypertension increased 2.10 fold (P=0.024) in GDM & hGDM groups compared to the IGG and NGG groups. In the Cape Town cohort, a third of women in the GDM group and three-quarters in the hGDM group progressed to T2DM at 5 years. CONCLUSIONS The NPRP categorization identifies four distinct risk clusters of glycemia in pregnancy which may aid better decision making in routine management, avoid potential over-diagnosis of women at lower risk of complications and assist with diabetes prevention in women at high-risk after an index pregnancy with GDM.