Objective: We investigated the diagnostic value of the visually assessed electroencephalogram (EEG) in patients with mild Alzheimer's disease (AD), using the grand total of EEG (GTE) score. Methods: Forty-nine non-demented control subjects with and without minimal cognitive impairment from the general population and 86 probable AD patients (NINCDS-ADRDA criteria), consecutively referred to a memory clinic, participated in this study. Results: Frequency of rhythmic background activity (P < 0.05), diffuse slow activity (P < 0.001), and reactivity of the rhythmic background activity (P < 0.001) were statistically significant related to the diagnosis control subject or AD patient, using logistic regression analysis with adjustment for age and sex. When these subscores were used to confirm the diagnosis of AD, thus at high specificity of 89.1% (GTE cut-off point of 3), the sensitivity was 44.6% and positive predictive value was 88.1%. Incremental ruling-in and ruling-out curves showed a maximum diagnostic gain of 38% for a positive test result at a prior probability ranging from 30 to 40%. At high pretest probability levels of 80-90%, the diagnostic gain for a positive test result was low, varying from 7 to 14%. Conclusion: In conclusion, the visually assessed EEG may give a clinically meaningful contribution to the diagnostic evaluation of AD when there is diagnostic doubt.