Aims/hypothesis The aim was to investigate the relationship between severe hypoglycaemia and cognitive impairment in older patients with diabetes. Methods A sample of 302 diabetic patients aged ≥70 years was assessed for dementia or cognitive impairment without dementia in 2001–2002 and a subsample of non-demented patients ( n  = 205) was followed to assess cognitive decline. A history of severe hypoglycaemia was determined from self-reports, physician assessments and records of health service use for hypoglycaemia (HSH). Prospective HSH was determined up to 2006. Data analysis, including multiple logistic and Cox regression models, was used to determine whether: (1) there were cross-sectional associations between hypoglycaemia and cognitive status, (2) historical hypoglycaemia predicted cognitive decline, and (3) baseline cognitive status predicted subsequent HSH. Results There were significant cross-sectional associations between both cognitive impairment and dementia and hypoglycaemia. Independent risk factors for future HSH included dementia (hazard ratio 3.00, 95% CI 1.06–8.48) and inability to self-manage medications (hazard ratio 4.17, 95% CI 1.43–12.13). However, there were no significant associations between historical hypoglycaemia, incident HSH and cognitive decline. Conclusions/interpretation Dementia is an important risk factor for hypoglycaemia requiring health service utilisation. We found no evidence that hypoglycaemia contributes to cognitive impairment in older patients with diabetes.