OBJECTIVE: To investigate whether different degrees of maternal glucose intolerance are associated with the risk of spontaneous preterm birth. METHODS: We performed a cohort study of 46,230 pregnancies screened by a 50-g, 1-hour oral glucose tolerance test between 24 and 28 gestation weeks at the Northern California Kaiser Permanente Medical Care Program. Spontaneous preterm birth was defined as an infant born at less than 37 gestation weeks with at least one of the following: spontaneous labor, preterm premature rupture of membranes, or incompetent cervix. Glucose tolerance status was categorized as normal screening (1-hour plasma glucose less than 140 mg/dL), abnormal screening (1-hour plasma glucose of at least 140 mg/dL with a normal diagnostic 100-g, 3-hour oral glucose tolerance test result), Carpenter-Coustan (plasma glucose measurements during the diagnostic oral glucose tolerance test met the thresholds but were lower than the National Diabetes Data Group thresholds), and gestational diabetes mellitus (GDM) by the National Diabetes Data Group criteria. RESULTS: One thousand nine hundred fifty-six spontaneous preterm births occurred. Age-adjusted incidences of spontaneous preterm birth were 4.0% in normal screening, 5.0% in abnormal screening, 6.7% in Carpenter-Coustan, and 6.7% in GDM. In a logistic regression model adjusted for age, race-ethnicity, preeclampsia-eclampsia-pregnancy-induced hypertension, chronic hypertension, polyhydramnios, and birth weight for gestational age, pregnancies with abnormal screening, Carpenter-Coustan, and GDM had a significantly higher risk of spontaneous preterm birth than pregnancies with normal screening (relative risk [95% confidence interval]: 1.23 [1.08, 1.41], 1.53 [1.16, 2.03], and 1.42 [1.15-1.77], respectively). CONCLUSION: The risk of spontaneous preterm birth increased with increasing levels of pregnancy glycemia. This association was independent of perinatal complications that could have triggered early delivery.