BACKGROUND: Associations between sleep duration and incident diabetic microvascular complications remain uncertain. The potential biological pathways underlying the association are unclear. METHODS: This prospective cohort study included 24,081 participants with type 2 diabetes from the UK Biobank. Habitual sleep duration was obtained via a baseline questionnaire. Associations of sleep duration with microvascular complications were analyzed using Cox regression models. RESULTS: The association of sleep duration with composite microvascular complications followed U-shaped patterns. Compared with sleeping for 7 hrs/day, the hazard ratios (95 % confidence intervals) of ≤ 5 and ≥ 10 hrs/day were 1.21 (1.09, 1.35) and 1.24 (1.09, 1.42) for the composite outcome, 1.20 (1.04, 1.38) and 1.30 (1.09, 1.54) for diabetic kidney disease, 1.61 (1.30, 2.00) and 1.35 (1.01, 1.79) for diabetic neuropathy, and 1.10 (0.93, 1.30) and 1.17 (0.95, 1.44) for diabetic retinopathy. Body mass index, gamma-glutamyl transferase, C-reactive protein, alkaline phosphatase, total cholesterol, and low-density lipoprotein cholesterol collectively explained 20.34 % of the association between sleep duration and composite microvascular complications. CONCLUSIONS: Our study demonstrated that both short and long sleep durations were independently associated with increased risk of diabetic microvascular complications. The associations were partially mediated by metabolic biomarkers related to obesity, systemic inflammation, liver function, and lipid profile.