We evaluated the therapeutic effects of non-pharmacological interventions (cognitive training, physical activity, and non-invasive brain stimulation) on cognitive symptoms in Parkinson's disease. A comprehensive literature search for non-pharmacological intervention randomized controlled trials was performed and effect sizes were calculated for each suitable study intervention approach and cognitive domain. Despite the heterogeneity of the study results, we report level B evidence for the probable efficacy of cognitive training in improving or maintaining attention/working memory and memory domains. Level C (possible efficacy) evidence was found for specific physical training types with respect to enhancing executive functions. Non-invasive brain stimulation techniques and combinatorial approaches show preliminary but promising results. Prediction markers evaluating distinct treatment responses should be identified that would help to choose the best candidates for specific treatment strategies and cognitive symptoms. Future directions and recommendations are discussed.