© 2018 American Medical Association. All rights reserved. IMPORTANCE Lactation duration has shown weak protective associations with incident diabetes (3%-15% lower incidence per year of lactation) in older women based solely on self-report of diabetes, studies initiated beyond the reproductive period are vulnerable to unmeasured confounding or reverse causation from antecedent biochemical risk status, perinatal outcomes, and behaviors across the childbearing years. OBJECTIVE To evaluate the association between lactation and progression to diabetes using biochemical testing both before and after pregnancy and accounting for prepregnancy cardiometabolic measures, gestational diabetes (GD), and lifestyle behaviors. DESIGN, SETTING, AND PARTICIPANTS For this US multicenter, community-based 30-year prospective cohort study, there were 1238 women from the Coronary Artery Risk Development in Young Adults (CARDIA) study of young black and white women ages 18 to 30 years without diabetes at baseline (1985-1986) who had 1 or more live births after baseline, reported lactation duration, and were screened for diabetes up to 7 times during 30 years after baseline (1986-2016). EXPOSURES Time-dependent lactation duration categories (none, >0 to 6 months, >6 to <12 months, and 12 months) across all births since baseline through 30 years. MAIN OUTCOMES AND MEASURES Diabetes incidence rates per 1000 person-years and adjusted relative hazards (RH) with corresponding 95% CIs, as well as proportional hazards regression models adjusted for biochemical, sociodemographic, and reproductive risk factors, as well as family history of diabetes, lifestyle, and weight change during follow-up. RESULTS Overall 1238 women were included in this analysis (mean [SD] age, 24.2 [3.7] years; 615 black women). There were 182 incident diabetes cases during 27 598 person-years for an overall incidence rate of 6.6 cases per 1000 person-years (95% CI, 5.6-7.6); and rates for women with GD and without GD were 18.0 (95% CI, 13.3-22.8) and 5.1 (95% CI, 4.2-6.0), respectively (P for difference < .001). Lactation duration showed a strong, graded inverse association with diabetes incidence: adjusted RH for more than 0 to 6 months, 0.75 (95% CI, 0.51-1.09); more than 6 months to less than 12 months, 0.52 (95% CI, 0.31-0.87), and 12 months or more 0.53 (0.29-0.98) vs none (0 days) (P for trend = .01). There was no evidence of effect modification by race, GD, or parity. CONCLUSIONS AND RELEVANCE This study provides longitudinal biochemical evidence that lactation duration is independently associated with lower incidence of diabetes. Further investigation is required to elucidate mechanisms that may explain this relationship.