Carsharing, an alternative to car ownership, is being encouraged by many national governments as a means to alleviate air pollution and traffic congestion. Previously, many carsharing companies determined service locations through trial and error, but they currently define their parking locations in metropolitan cities for maximum customer coverage. However, identifying carsharing locations according to the experiences of the pioneering cities might not yield valid results in some Asian countries where carsharing systems are unknown. Hence, this study examines the characteristics of carsharing users in Daejeon, a small Korean city, to determine that city's optimal carsharing service locations. A geographic information system was used to analyze and determine the best spatial areas according to two data categories: internal and external demand factors. Suitable carsharing locations were ranked by the results of a grid analysis. Thirty optimal locations were then determined from the location-allocation model in a network analysis module. Determining optimal carsharing locations should also be directly correlated with the reduction of carbon dioxide emissions. Carbon dioxide emission reduction from carsharing was predicted at 62,070 tCO2eq for the year 2013; emission reductions were predicted to increase further to 172,923 tCO2eq by 2020. Thus, carsharing is an innovative strategy for traffic demand management that can alleviate air pollution. The results of this study indicate that further research is necessary to examine the relationship between optimal carsharing locations and carbon dioxide emission reduction from using lower-emission carsharing vehicles, such as electric vehicles.