Recent randomized control trials and meta-analyses of experimental studies indicate positive effects of non-pharmacological cognitive training on the cognitive function of healthy older adults. Furthermore, a large-scale randomized control trial with older adults, independent at entry, indicated that training delayed their cognitive and functional decline over a five-year follow-up. This supports cognitive training as a potentially efficient method to postpone cognitive decline in persons with mild cognitive impairment (MCI). Most of the research on the effect of cognitive training in MCI has reported increased performance following training on objective measures of memory whereas a minority reported no effect of training on objective cognitive measures. Interestingly, some of the studies that reported a positive effect of cognitive training in persons with MCI have observed large to moderate effect size. However, all of these studies have limited power and few have used long-term follow-ups or functional impact measures. Overall, this review highlights a need for a well-controlled randomized trial to assess the efficacy of cognitive training in MCI. It also raises a number of unresolved issues including proper outcome measures, issues of generalization and choice of intervention format.