Considering the fast development of urban areas and the increasing environmental awareness, we focus in this paper on a taxi-sharing system for improving the efficiency of public transport. In particular, we present a realistic and large-scale evaluation of the environmental impact of this system. The proposed method improves the efficiency of this form of public transport by determining, on the fly, the best match between taxi user requests, which leads to a reduction of the total and vacant travel distance. Our results show important reductions of CO2 emissions with full deployment of the system, reaching 9% under a higher demand scenario.