Aims/hypothesis: There are a number of glycemic definitions for prediabetes; however, the heterogeneity in diabetes transition rates from prediabetes across different glycemic definitions in major US cohorts has been unexplored. We hypothesize that a significant source of variation in the transition rate are cohorts themselves. We estimate the variability in risk and relative risk of diabetes based on diagnostic criteria like fasting glucose and hemoglobin A1C% (HbA1c%). Methods: We estimated transition rate from prediabetes, as defined by fasting glucose between 100-125 and/or 110-125 mg/dL, and HbA1c% between 5.7-6.5% in participant data from the Framingham Heart Study (FHS) Generation 2, FHS Generation 3, Multi-Ethnic Study on Atherosclerosis, Atherosclerosis Risk in Communities, and the Jackson Heart Study. We estimated the heterogeneity and prediction interval across cohorts, stratifying by age, sex, and body mass index. Among individuals with prediabetes, we estimated the relative risk for obesity, blood pressure, education, age, and sex for diabetes. Results: There is substantial heterogeneity in diabetes transition rates across cohorts and prediabetes definitions with large prediction intervals. We observed the individuals with fasting glucose of 100-125 range from 4-14% per 100 person years and 110-125 mg/dL ranging from 2-18 per 100 person-years. For HbA1C between 5.7-6.5%, the transition rate ranged from 2.5-11 per 100 person years (I2 for heterogeneity was greater than 93% for all definitions). Obesity and hypertension did not explain the differences in risk. Conclusion: The absolute transition rate from prediabetes to diabetes significantly depends on both cohort and prediabetes definitions.