Mental capacity was assessed in 311 apparently self-sufficient patients (> or = 60 years of age, 170 men) under stabilised oral anticoagulant treatment (OAT) by administering the Hodkinson's Abbreviated Mental Test (AMT). The international normalized ratios (INR) recorded during the 3 months before and the 3 after the data of test administration were examined by the INR-Day software program. The percentage of time spent within, below or above the intended therapeutic range was calculated in patients who scored abnormally at AMT, and compared with matched controls with normal AMT results. Forty patients [12.9%; 28 women (19.8%) and 12 men (7.1%), P < 0.0011] had abnormal AMT results; the rate seemed to increase with age. Most of these patients (35, 75%) had only elementary education. Patients with abnormal AMT results spent more time outside the intended therapeutic ranges than 40 matched controls (20.9% of the observed time vs 13.7%, P < 0.0001; odds ratio 1.68, CI 1.53-1.84). Unsuspected reduction of mental ability or attention levels was found in a number of elderly patients receiving OAT; these patients presented longer periods of either under- or over-anticoagulation and were, therefore, exposed to a higher risk of thrombotic or bleeding complications. Anticoagulation clinics would be advised to assess mental abilities in elderly patients before starting OAT.