OBJECTIVE: In the U.K., one-third of patients receiving treatment with dialysis have diabetes. Guidelines from organizations representing patients with renal disease or diabetes advocate tight glycemic control in patients with end-stage renal disease, despite glucose-lowering trials having excluded these patients. RESEARCH DESIGN AND METHODS: Using national U.K. Renal Registry data, we tested whether glycemia as measured by hemoglobin (Hb) A(1c) (HbA(1c)) level is associated with death in adults with diabetes starting hemodialysis or peritoneal dialysis between 1997 and 2006, and observed for at least 6 months. Of 7,814 patients, we excluded those who had died within 6 months; had received transplants; were lost/recovered; or lacked measures of HbA1c, ethnicity, or Hb. Categorizing HbA1c measured in the first 6 months of starting dialysis as <6.5% (<48 mmol/mol), 6.5-7.4% (48-57 mmol/mol) (reference value), 7.5-8.4% (58-68 mmol/mol), and ≥8.5% (≥69 mmol/mol), we adjusted in proportional hazards models for age, sex, ethnicity, deprivation, year, dialysis type, and Hb, and tested for interactions. RESULTS: Of 3,157 patients observed for a median time of 2.7 years, 1,688 died. For patients ≥60 years of age, we found no association between HbA1c and death; among younger patients, relative to those with HbA(1c) values 6.5-7.4%, the hazard ratio for HbA(1c) level 7.5-8.4% was 1.2 (95% CI 0.9-1.5), and for HbA(1c) level >8.5% was 1.5 (1.2-1.9). The projected difference in median survival time between younger patients with a reference HbA1c value versus >8.5% was 1 year. CONCLUSIONS: In the absence of trials, and confounding notwithstanding, these observational data support improved glycemic control in younger patients prior to and during dialysis.