OBJECTIVE: To identify the range of glycemic levels associated with the lowest rates of complications and mortality in older diabetic patients. RESEARCH DESIGN AND METHODS: We conducted a retrospective cohort study (2004-2008) of 71,092 patients with type 2 diabetes, aged >/=60 years, enrolled in Kaiser Permanente Northern California. We specified Cox proportional hazards models to evaluate the relationships between baseline glycated hemoglobin (A1C) and subsequent outcomes (nonfatal complications [acute metabolic, microvascular, and cardiovascular events] and mortality). RESULTS: The cohort (aged 71.0 +/- 7.4 years [means +/- SD]) had a mean A1C of 7.0 +/- 1.2%. The risk of any nonfatal complication rose monotonically for levels of A1C >6.0% (e.g., adjusted hazard ratio 1.09 [95% CI 1.02-1.16] for A1C 6.0-6.9% and 1.86 [1.63-2.13] for A1C >/=11.0%). Mortality had a U-shaped relationship with A1C. Compared with the risk with A1C <6.0%, mortality risk was lower for A1C levels between 6.0 and 9.0% (e.g., 0.83 [0.76-0.90] for A1C 7.0-7.9%) and higher at A1C >/=11.0% (1.31 [1.09-1.57]). Risk of any end point (complication or death) became significantly higher at A1C >/=8.0%. Patterns generally were consistent across age-groups (60-69, 70-79, and >/=80 years). CONCLUSIONS: Observed relationships between A1C and combined end points support setting a target of A1C <8.0% for older patients, with the caution that A1Cs <6.0% were associated with increased mortality risk. Additional research is needed to evaluate the low A1C-mortality relationship, as well as protocols for individualizing diabetes care.