OBJECTIVE: Multiple lines of evidence support a complex relationship between obstructive sleep apnea (OSA) and diabetes. However, no population-based study has evaluated the potential bidirectional association between these two highly prevalent disorders. RESEARCH DESIGN AND METHODS: We followed 146,519 participants from the Nurses' Health Study (NHS; 2002-2012), Nurses' Health Study II (NHSII; 1995-2013), and Health Professionals Follow-up Study (HPFS; 1996-2012) who were free of diabetes, cardiovascular disease, and cancer at baseline. Cox proportional hazards models were used to estimate hazard ratios (HRs) for developing diabetes according to OSA status. In parallel, we used similar approaches to estimate risk of developing OSA according to diabetes status among 151,194 participants free of OSA, cardiovascular disease, and cancer at baseline. In all three cohorts, diagnoses of diabetes and OSA were identified by validated self-reports. RESULTS: Similar results were observed across the three cohorts. In the pooled analysis, 9,029 incident diabetes cases were identified during follow-up. After accounting for potential confounders, the HR (95% CI) for diabetes was 2.06 (1.86, 2.28) comparing those with versus without OSA. The association was attenuated but remained statistically significant after further adjusting for waist circumference and BMI (HR 1.37 [95% CI 1.24, 1.53]), with the highest diabetes risk observed for OSA concomitant with sleepiness (1.78 [1.13, 2.82]). In the second analysis, we documented 9,364 incident OSA cases during follow-up. Compared with those without diabetes, the multivariable HR (95% CI) for OSA was 1.53 (1.32, 1.77) in individuals with diabetes. Adjustment for BMI and waist circumference attenuated the association (1.08 [1.00, 1.16]); however, an increased risk was observed among those with diabetes who used insulin compared with those without diabetes (1.43 [1.11, 1.83]), particularly among women (1.60 [1.34, 1.89]). CONCLUSIONS: OSA is independently associated with an increased risk of diabetes, whereas insulin-treated diabetes is independently associated with a higher risk of OSA, particularly in women. Clinical awareness of this bidirectional association may improve prevention and treatment of both diseases. Future research aimed at elucidating the mechanisms that underlie each association may identify novel intervention targets.