This study aimed to evaluate the inter-rater reliability of streetscape audits among online observations using the Microscale Audit of Pedestrian Streetscapes-Global version (MAPS-Global) in Japan. MAPS-Global observations were conducted on routes with distances ranging from 400 to 725 m from a residence toward a non-residential destination. Google Street View audits were independently conducted by two trained raters on each route. A tiered scoring system was applied to summarize the items at multiple levels of aggregation. Positive and negative valence scores were created based on the expected association with physical activity. Inter-rater reliability analyses were performed using kappa statistics or intraclass correlation coefficients (ICC). Of the 32 older adults participating in an intervention study in the community-wide physical activity promotion project in Fujisawa City, 19 addresses were used, excluding those with nearby addresses. Results demonstrated “excellent” agreement for most of the summary scores analyzed (kappa or ICC values of 0.75 or higher [80.4 %]), while 6.5 % of items exhibited “good” agreement (ICC = 0.60–0.74). By contrast, only 13.0 % of the scales had ICC values lower than 0.60 (“fair” or “poor” reliability). The results illustrated high reliability for the grand summary scores and composite subscale measures. However, caution should be exercised when interpreting subscale scores for less frequently observed negative attributes and aesthetic/social characteristics. The results presented in this study support the application of online observations using MAPS-Global in urban areas of Japan, which could be implemented to inform decisions related not only to physical activity but also to traffic safety.