OBJECTIVE Poor sleep has been identified as a risk factor for poor glycemic control in individuals with type 2 diabetes (T2D). As optimal sleep can be characterized in several ways, we evaluated which sleep characteristics are most strongly associated with glycated hemoglobin A1c (HbA1c). RESEARCH DESIGN AND METHODS A total of 172 patients with T2D completed 7-day wrist-actigraphy and sleep questionnaires. Linear regression was used to evaluate associations between sleep measures (total sleep duration, variability in sleep duration, midsleep time, variability in midsleep time, sleep efficiency, subjective sleep quality, and subjective insomnia symptoms) and HbA1c, individually and in concert. RESULTS Variability in sleep duration was individually most strongly associated with HbA1c (b 5 0.239; P 5 0.002; R2 5 4.9%), followed by total sleep duration (U-shaped: b 5 1.161/b2 5 1.044; P 5 0.017/0.032; R2 5 4.3%), subjective sleep quality (b 5 0.191; P 5 0.012; R2 5 3.6%), variability in midsleep time (b 5 0.184; P 5 0.016; R2 5 3.4%), and sleep efficiency (b 5 20.150; R2 5 2.3%). Midsleep time and subjective insomnia symptoms were not associated with HbA1c. In combination, variability in sleep duration, total sleep duration, and subjective sleep quality were significantly associated with HbA1c, together explaining 10.3% of the variance in HbA1c. Analyses adjusted for covariates provided similar results, although the strength of associations was generally decreased and showing total sleep duration and subjective sleep quality to be most strongly associated with HbA1c, together explaining 6.0% of the variance in HbA1c. CONCLUSIONS Sleep in general may be a modifiable factor of importance for patients with T2D. The prevention of sleep curtailment may serve as a primary focus in the sleep-centered management of T2D.