OBJECTIVES: This study performed meta-analysis on the literatures that surveyed cognitive test to forecast unsafe driving by older drivers and identified objective and consistent cognitive test for predicting unsafe driving of older drivers. SELECTION CRITERIA: The study of RCT (Randomized Control Trial) that conducted cognitive test by classifying older drivers into safe-drivers and unsafe-drivers was done and a total of nine studies suitable for the selection criteria were chosen. SEARCH STRATEGY: To meet subject selection, online search was performed by keyword such as "Older", "Driving", "Safe", "Cognition", etc. Qualitative analysis of the study was conducted using Jadad evaluation. Quantitative analysis also conducted statistical heterogeneity, effect size, sensitivity and publication bias every cognitive assessment tool. RESULT: The Jadad evaluation grade of the studies was assessed on papers of a high quality -all study received over 3 points. The result of the effect sizes was that TMT-B, TMT-A, UFOV-subtest 2 and MMSE were statistically significant (P < .05). As a result, TMT-B was "Big effect size", TMT-A and UFOV-subtest 2 were "Medium effect sizes" and MMSE was "Small effect size." CONCLUSION: TMT-A, TMT-B and UFOV-subtest 2 were found to be useful as cognitive test tools to forecast unsafe driving of older drivers.