Continuous ambient air monitoring systems have been introduced worldwide. However, such monitoring forces autonomous communities to bear a significant financial burden. Thus, it is important to identify pollutant-monitoring stations that are less efficient, while minimizing loss of data quality and mitigating effects on the determination of spatiotemporal trends of pollutants. This study describes a procedure for optimizing a constant ambient air monitoring system in the Kanto region of Japan. Constant ambient air monitoring stations in the area were topologically classified into four groups by cluster analysis and principle component analysis. Then, air pollution characteristics in each area were reviewed using concentration contour maps and average pollution concentrations. We then introduced three simple criteria to reduce the number of monitoring stations: (1) retain the monitoring station if there were similarities between its data and average data of the group to which it belongs; (2) retain the station if its data showed higher concentrations; and (3) retain the station if the monitored concentration levels had an increasing trend. With this procedure, the total number of air monitoring stations in suburban and urban areas was reduced by 36.5%. The introduction of three new types of monitoring stations is proposed, namely, mobile, for local non-methane hydrocarbon pollution, and Ox-prioritized.