Research has been showing that oculomotor measurements can accurately detect higher-order cognitive deficits, namely in memory and visuospatial processing deficits (Rizzo et al., 2004; Garbutt et al., 2008; Crutcher et al., 2009). Literature has still failed to analyze the impact of cognitive effort in the eye movement behavior as a tool to distinguish healthy subjects from those with cognitive decline due to Alzheimer's disease. 26 age and education matched participants (6 AD, 10 MCI, 10 NC) were submitted to a computational task composed by a familiarization phase and a test phase. During the familiarization phase, participants were instructed to look at two initial images. After that, there was a variable time delay of 5 or 90 seconds. During the test phase, other two images were presented, one of them was identical to previously seen, the other was a new one. Subjects were instructed to select the novel image which was either similar or non similar to the familiar stimulus. Eye-movement bahavior was analyzed based on four main parameters: number of visits (VC), duration of visits (VD) number of fixations (FC), fixation duration (FD). Parameters were analyzed regarding diagnostic, stimuli similarity and time between trials. Stimuli similarity revealed significant interference in the following parameters: VC, PrFC, PrVD, PrVC. Stimuli similarity also had impact on eye-movement pattern when associated with time delay between the familiarization phase and the test phase (PrFD, PrFC, PrVC). Both the proportion of time fixated on the novel stimulus (PrFD) and the proportion of the number of visits on the novel stimulus (PrVC) were affected by the participants’ diagnosis, when combined with stimuli similarity and time delay, respectively.