County-level highway, as a transport carrier to expand the external economic exchanges of rural areas, plays an important role as a hub in the rural road network system, and it is an important infrastructure for the country to promote the construction of new urbanization. In order to ensure the sustainable development of county-level highway, according to the relevant policies in China, the research on the investment effect evaluation of county-level highway projects has been regarded as a focus in the “14th Five-Year Plan”. Therefore, evaluating the investment benefits of the in-service county-level highways by identifying key evaluation factors has become an important research topic for new urbanization development. Since there is a lack of a comprehensive and reasonable evaluation index system. Therefore, this paper has taken the policy applicability and maximization of comprehensive benefits as the research goal, and started from five criteria, including new urbanization development ability, road network construction ability, economic benefits, social benefits and environmental benefits. Later, 14 second-level factors including industrial development ability, coverage of road network structure and national economic evaluation conclusion, etc., as well as a total of 58 third-level factors including the growth rate of the number of townships service industries and service industries, the increase of highway transport tools and regional urbanization growth rate, etc. were screened and constructed as the investment effect evaluation system. Subsequently, according to the characteristics of gray fuzzy of county-level highway indexes, this paper established a hybrid model applicable to the multi-attribute investment effect evaluation of new urbanization county-level highway, which used the game combination weighting method (GCWM), to combine the entropy power method and hierarchical analysis method to assign weights and reasonably calculate the comprehensive weights of indexes, and on this basis, it combined grey Euclid relational analysis method (GERAM) and the attribute theory to rank the correlation degrees of evaluation objects and further classify the investment effect level of the project. Finally, the hybrid model was applied to the county-level highway under the new urbanization construction demonstration counties to derive the optimal solution, as well as the optimization paths and measures for other auxiliary solutions. The results of this study provide an effective theoretical basis for the evaluation of the investment yield of rural transportation infrastructure with the support of policies, and provide a reliable practical method for the developing countries to achieve the maximum comprehensive benefits of the county-level highway investment programs with limited funds.