Study Objectives: Studies examining the longitudinal association of untreated obstructive sleep apnea (OSA) with diabetes in population samples are limited. This study therefore examined the relationship between previously undiagnosed OSA with incident type 2 diabetes in community-dwelling men aged ≥ 40 y. Methods: The Men Androgen Inflammation Lifestyle Environment and Stress (MAILES) Study is a longitudinal population-based cohort in Adelaide, South Australia. Clinic assessments at baseline and follow-up identified diabetes (self-reported doctor diagnosed, fasting plasma glucose ≥ 7.0 mmol/L, glycated hemoglobin ≥ 6.5% or diabetes medication use) and included anthropometry. At cohort follow-up (2010-2012), n = 837 underwent full in-home unattended polysomnography (PSG, Embletta X100, Broomfield, CO). Results: Of 736 men free of diabetes at baseline, incident diabetes occurred in 66 (9.0%) over a mean follow-up time of 56 mo (standard deviation = 5, range: 48-74 mo). Incident diabetes was associated with current oxygen desaturation index (3%) ≥ 16 events/h (odds ratio [OR]: 1.85 [1.06-3.21]), and severe OSA [OR: 2.6 (1.1-6.1)], in adjusted models including age, percentage total body fat, and weight gain (> 5 cm waist circumference). An age-adjusted association of incident diabetes with percentage of total sleep time with oxygen saturation < 90% did not persist after adjustment for percentage of body fat. No modification of these relationships by excessive daytime sleepiness was observed. Conclusions: Severe undiagnosed OSA and nocturnal hypoxemia were independently associated with the development of diabetes. A reduction in the burden of undiagnosed OSA and undiagnosed diabetes is likely to occur if patients presenting with one disorder are assessed for the other.