Diabetes is associated with increased mortality in older adults, but the specific contributions of diabetes-associated clinical conditions and of increasing hyperglycaemia to mortality risk are unknown. We evaluated whether cardiovascular disease, comorbidities, or degree of hyperglycaemia, particularly severe hyperglycaemia, affected diabetes-related mortality risk in older, disabled women. Six-year mortality follow-up of a random sample of 576 disabled women (aged 65-101 years), recruited from the Medicare eligibility list in Baltimore (MD, USA). All-cause and cardiovascular mortality were evaluated by diabetes status: no diabetes; diabetes with mild, moderate, and severe hyperglycaemia [defined by tertiles of glycosylated haemoglobin (GHB) among women with diabetes]. Diabetes with mild, moderate, and severe hyperglycaemia was associated with an increased hazard rate (HR) for all-cause mortality, even after adjustment for demographics, risks for cardiovascular disease, cardiovascular and non-cardiovascular conditions, and other known mortality risks. A dose-response effect was suggested [mild hyperglycaemia, HR 1.81, 95% confidence interval (CI) 1.03, 3.17; moderate hyperglycaemia, HR 2.02, 95% CI 1.34, 3.57; severe hyperglycaemia, HR 2.22, 95% CI 1.17, 4.25]. Women with diabetes had a significantly increased HR for non-cardiovascular death, but not for cardiovascular death, compared with those without diabetes. Diabetes, whether characterized by mild, moderate or severe hyperglycaemia, appears to be an independent risk factor for excess mortality in older disabled women and this risk may increase with increasing hyperglycaemia. This mortality risk is not completely explained by vascular complications, and involves non-cardiovascular deaths. Risks and benefits of diabetes management, including glycaemic control and management of vascular and other comorbidities, should be studied in older people with complications and comorbidities.