© 2015, Springer-Verlag Berlin Heidelberg. Aims/hypothesis: The impact of marijuana use on metabolic health is largely unknown. This study sought to clarify the cross-sectional and longitudinal associations between self-reported marijuana use, and prediabetes (defined as fasting glucose 5.6–6.9 mmol/l, 2 h glucose post OGTT 7.8–11.0 mmol/l or HbA1c 5.7–6.4% [39–47 mmol/mol]) and diabetes. Methods: Data from the community-based Coronary Artery Risk Development in Young Adults (CARDIA) study were used to determine marijuana use and the presence of prediabetes and diabetes among participants. The association between marijuana use and the prevalence of prediabetes and diabetes was examined in 3,034 participants at CARDIA examination year 25 (2010–2011), while the incidence of prediabetes and diabetes according to previous marijuana use was assessed in 3,151 individuals who were free from prediabetes/diabetes at year 7 (1992–1993) and who returned for at least one of the four subsequent follow-up examinations over 18 years. Results: The percentage of individuals who self-reported current use of marijuana declined over the course of the study’s follow-up. After multivariable adjustment, higher odds of prediabetes were found for individuals who reported current use of marijuana (OR 1.65 [95% CI 1.15, 2.38]) and a lifetime use of 100 times or more (OR 1.49 [95% CI 1.06, 2.11]), compared with individuals who reported never using marijuana. There was no association between marijuana use and diabetes at CARDIA examination year 25. Over 18 years of follow-up, a greater risk of prediabetes (but not diabetes) was found for individuals who reported a lifetime use of marijuana of 100 times or more (HR 1.39 [95% CI 1.13, 1.71]), compared with individuals who had never used marijuana. Conclusions/interpretation: Marijuana use in young adulthood is associated with an increased risk of prediabetes by middle adulthood, but not with the development of diabetes by this age.