Neuropsychological function was compared in three well-matched groups of subjects: Group 1, 20 diabetic patients with hypertension, mean age 69.1 +/- 4.8 years, 14 males and 6 females; Group 2, 20 normotensive diabetic patients, mean age 69.0 +/- 6.2 years, 14 males and 6 females; Group 3, 20 healthy community controls, mean age 68.1 +/- 4.5 years, 13 males and 7 females. There were no significant differences between the groups in education or estimated IQ using the NART (National Adult Reading Test). Groups 1 and 2 did not differ significantly in duration of diabetes (mean 10.6 and 9.5 years, respectively), or mean glycosylated haemoglobin, HbA1 (mean 9.8 and 10.6%, respectively), or mean blood glucose before and after testing. On a battery neuropsychological tests, sensitive to cognitive impairment in older subjects, analysis of covariance using estimated IQ as the covariate showed no significant differences between the groups on tests of recall, with (Brown-Peterson Test) and without (Kendrick Object Learning Test) interference, forward and backward digit span, concentration (serial subtraction), verbal fluency, immediate and delayed prose recall, digit symbol substitution or psychomotor speed (Kendrick Digit Copying Test). These results provide no support for an association between cognitive deficits and Type 2 diabetes mellitus in older subjects or for the view that such deficits may also be mediated by hypertension.