High-resolution modeling approach is increasingly being considered as a necessary step for improving the monitoring and predictions of regional air quality. This is especially true for highly urbanized region with complex terrain and land-use. This study uses Community Multiscale Air Quality (CMAQ) model coupled with MM5 mesoscale model for a comprehensive analysis to assess the suitability of such high-resolution modeling system in predicting ozone air quality in the complex terrains of Osaka, Japan. The 1-km and 3-km grid domains were nested inside a 9-km domain and the domain with 1-km grid covered the Osaka region. High-resolution Grid Point Value-Mesoscale Model (GPV-MSM) data were used after suitable validation. The simulated ozone concentrations were validated and evaluated using statistical metrics using performance criteria set for ozone. Daily maxima of ozone were found better simulated by the 1-km grid domain than the coarser 9-km and 3-km domains, with the maximum improvement in the mean absolute gross error about 3 ppbv. In addition, 1-km grid results fared better than other grids at most of the observation stations that showed noticeable differences in gross error as well as correlation. These results amply justify the use of the integrated high-resolution MMS-CMAQ modeling system in the highly urbanized region, such as the Osaka region, which has complex terrain and land-use.