In recent years, quantitative variables derived from the electroencephalogram (EEG) attract an increasing interest for the evaluation of neurodegenerative diseases, as EEG registers the neuro-electric activity with a high temporal resolution and provides a cost-effective and easily accessible, non-invasive method. Event-related oscillations (EROs) as oscillatory responses in the EEG to specific events further provide the possibility to track the cognitive decline in a task-specific manner. Current study in search for potential ERO biomarkers to distinguish different stages of cognitive decline along the Alzheimer's Disease (AD) continuum re-analyzed a combined set of data collected and analyzed in previous studies by Basar and coworkers. Target responses of a visual oddball experiment recorded from 33 AD patients, 46 Mild Cognitive Impairment (MCI) patients and 48 age, gender, and education matched normal elderly controls were analyzed for both evoked (phase-locked) and total (phase-locked + non-phase-locked) ERO powers in delta, theta, alpha, beta and gamma bands by applying continuous wavelet transform (WT) on averaged and single trial data, respectively. The cluster-based non-parametric permutation test implemented in the FieldTrip toolbox revealed significant differences among the three groups. While the total delta and theta responses already significantly declined in the MCI stage with further spatial expansion of the decline in AD, the evoked delta response reached a statistically significant reduction level in the AD stage. We obtained no significant difference among groups for alpha, beta and gamma frequency bands. These results suggest that total delta and theta EROs to oddball targets may be useful for early detection of the disease in MCI stage, while the evoked delta response allows detecting the conversion to AD.