Combining a strong economic performance and a sustainable use of natural resources is one of the major objectives of current Chinese agricultural policy. Agricultural eco-efficiency facilitates assessing sustainable development of agriculture in an integrated way, thus providing good guidance for decision making. For panel data from 2003 through 2010, a valuation approach-based on a capital analysis and the concept of opportunity costs-is used to assess agricultural eco-efficiency in China. In addition, benchmarking is an important component of the approach, because it reflects a judgment in determining the cost of all capital forms. These capital forms are land, labor, agricultural water consumption, chemical fertilizer, plastic film, pesticide, diesel, gross power of agricultural machinery, chemical oxygen demand (COD) and total nitrogen. Two benchmarks were chosen, the weighted average return on capital and the best economic performance of each capital form. Depending on the applied benchmark, it was found that the two measures of eco-efficiencies differed, but the variations were similar year to year. Because the main objective of this paper is to understand why evaluation objects differ in their eco-efficiency, our assessment adopted the first benchmark. A regression model was then introduced to analyze the determinants of eco-efficiency. The key results are as follows: (1) Using the weighted average return on capital form as a benchmark, the average ecological value is negative in 2003, increasing to a maximum in 2010. The value went from negative to positive in 2007. Among the ten capital forms considered, labor capital and COD capital proved to be the critical resources and drivers for value creation in different years. Agricultural eco-efficiency in China is low, but recorded significant progress over the observed period. A low average eco-efficiency of 0.7 was calculated in 2003, increasing to almost 1.34 in 2010. (2) Spatial distribution characteristics are also significant. The agro-ecological areas in the Northeast Zone, Loess Plateau Zone and Huang-Huai-Hai Plain Zone, which are the traditional major grain-producing areas in China, have a low rate of eco-efficiency. Provinces located north of the Qinling Mountains-the Huaihe River line, comprising the Heilongjiang, Shanxi, Ningxia, Gansu and Xinjiang main grain-producing areas, recorded lower values of eco-efficiency in 2010 than in 2003. Conversely, southern provinces such as Sichuan and Hainan had a high eco-efficiency rate. The reasons behind these regional differences are explained in the study. (3) Taking into account the resource endowment of agricultural production, the characteristics of labor, the input of resources, and the effects of agricultural policy, our empirical model shows that both structural and managerial characteristics have an impact on eco-efficiency. For example, cultivation area per crop and education level of farmers have a significant positive effect on eco-efficiency. A managerial characteristic is chemical fertilizer application per area. More chemical fertilizer application per area leads to a lower eco-efficiency rating in China. The gross power of agricultural machinery, agricultural water consumption, and subsidy levels show similar results as chemical fertilizer application per area. Overall, the study is useful in its identification of weak points in agriculture development in China, information that may support future development of relevant policies and government interventions.