Mild cognitive impairment is sometimes regarded as related to aging. However, statistically every second case turns into full dementia, which still is resistant to any treatment. It is therefore desirable to recognize deviations from normality as early as possible. This might be feasible by using quantitative EEG analysis in the presence of mental work. The present retrospective data analysis revealed a new quantitative biomarker indicating the degree of impairment. Current source density was calculated from 16 channel EEG using CATEEM ® software. Four different conditions were analyzed: relaxed state, performing a d2-concentration test, a calculation performance test and a memory test for 5 min each. Subjects older than 40 years were divided into two groups according to their DemTect score: 13 -18 (HC; n = 44) or 8 -12 (MCI; n = 45). Spectral power was chopped into six frequency ranges (delta, theta, alpha 1, alpha 2, beta 1 and beta 2). Average spectral power was enhanced in the MCI group in comparison to healthy subjects with respect to delta (p = 0.05) during relaxed state when all electrode positions were regarded. With respect to EEG recording during performance of three different psychometric tests it was recognized that mainly spectral changes during performance of the d2-concentration test were related to mild cognitive impairment. With regard to all electrode positions statistically significantly lower spectral power values were reached during the d2-test for delta (p = 0.001), theta (p = 0.0001) and alpha 1 waves (p = 0.08) in impaired subjects in comparison to healthy subjects. Regarding regions of interest increases of delta and theta power were seen in the fronto-temporal brain during performance of the d2-concentration test. These increases disappeared when looking at MCI data. In the centro-parietal region decreases of alpha and beta 1 power emerged, which were even larger in MCI subjects. No MCI-dependent changes were observed in the other two tests. A correlation was found between psychometric performance of the d2-test and the DemTect score (r = 0.51). MCI subjects had statistically significant worse performance in all three mental challenges in comparison to healthy volunteers. It is concluded that MCI can be characterized at an early stage by EEG recording in the relaxed state. High spectral delta and theta power in general and specifically at fronto-