Differentiating early Alzheimer's disease (AD) from depression with cognitive impairment is challenging in the elderly. To develop a model for differentiating these two conditions using electroencephalography (EEG), we enrolled 11 patients with early probable AD and 11 age- and cognitive function-matched patients with major depressive disorder (MDD) and compared the EEG relative powers of 9 scalp regions. Compared to the MDD group, the AD group had a higher global theta relative power (p=0.021). In the MDD group, beta relative power was higher in the mid-central region than in the left or right central regions (p<0.01). The prediction model that included global theta relative power and regional beta index was able to discriminate AD from MDD (AUC=0.893, p=0.002). A combination of global theta relative power and intra-individual regional differences in beta may differentiate early AD from MDD with cognitive impairment.