Dementia is a highly invalidating condition and, given the progressive aging of the population, one of the major issues that health systems will have to face in future years. Recently there has been an increase in the potential of diagnostic tools and pharmacological treatments for dementia; moreover, considerable interest has been expressed regarding non pharmacological interventions. However, the current evidence in support of non pharmacological treatments in patients affected by dementia still does not allow to draw definitive conclusions on what is the most effective treatment to apply, largely because of methodological difficulties and limitations of the studies so far carried out due to the complex nature of the disease. To address this need, we carried out a single blind randomized controlled study on the efficacy of computer cognitive rehabilitation in patients with mild cognitive decline. We here present preliminary data on 11 patients with diagnosis of Alzheimer's Disease (AD) and mild cognitive decline randomly assigned to treatment (a) or control (b) condition (i.e. specific vs. aspecific treatment). The specific treatment (a) consisted in a cycle of 12 individual sessions of computer exercises, while the control condition (b) consisted in sessions of semi-structured interviews with patients, conducted with the same frequency and time period as (a). Cognitive, behavioural and functional assessment was performed by an expert evaluator, blinded to the patients' group allocation. Preliminary results show a significant performance decline only in the control group at the 9-month follow-up compared to both baseline and the 3-month follow-up. Our results suggest that computer based cognitive training in patients with AD and mild cognitive decline is effective at least in delaying the continuous progression of cognitive impairment in AD.