Objective : The purpose of this study is to verify the quality level of treatment effect by performing a quality evaluation on the meta-analysis study of non-pharmacological interventions for dementia, and to present convergence research insights on the qualitative improvement of evidence-based practice in the domestic non-pharmacological intervention field in the future. Methods : For literature search, 'dementia' and 'meta-analysis' were selected, and KMbase, KoreaMed, RISS, KISS, and NDLS were used. Methodological quality of meta-analysis papers was analyzed according to the criteria of the AMSTAR evaluation tool. Results : The average score of methodological quality evaluation of meta-analysis papers was 6.16 (SD=1.90), which was average. Looking at the studies individually, there were 17 studies with a moderate level (points 5-8), two studies with a low level (points 0-4), and none of them had a high level (points 9-11). Of the 19 studies, 17 were published in KCI, and 5 Cognitive-based intervention, 4 physical activity intervention , 4 music therapy , 2 aroma therapy, and 1 virtual reality-based intervention, art therapy, recall therapy, and occupational therapy were surveyed. Looking at the frequency of use of the quality evaluation tool, the majority of tools was used ROB (59%), and the rest of tools was used SIGN (12%), PEDro score (12%), RoBANS (12%), and EQI (5%). Conclusion : Meta-analysis research in the field of dementia must be accompanied by the selection of literature to be included in the study and the qualitative verification process of the included research, and the overall research design and process needs to be systematically and rigorously conducted compared to previous methods.