As is universally recognized, the eco-efficiency is an important index to evaluate urban ecological civilization construction, and the Yellow River Basin is the main battlefield of national ecological civilization construction. Based on the panel data of the input, expected output and unexpected output of 43 cities along the Yellow River basin from 2007 to 2017, this paper utilizes super-efficiency Data Envelopment Analysis (DEA) based on relaxation variables to measure its eco-efficiency, also discusses the spatial and temporal variation pattern via integrating spatial auto-correlation and other methods, and empirically analyzes the spatial-temporal dynamic evolution and driving factors of the eco-efficiency of cities along the river basin. From 2007 to 2017, the overall eco-efficiency of 43 cities showed a trend of decline first and then rise. Most of the cities had a relatively low level of eco-efficiency and still had a large room for improvement. From the perspective of watershed division, the spatial pattern of ecological efficiency in upper, middle, and lower cities generally shows that the ecological efficiency of middle reaches is higher than that in upper and lower reaches. The Moran's I of the urban eco-efficiency in the study area in the past 10 years shows that the eco-efficiency first presents a spatial negative correlation and then evolves into a spatial positive correlation, and the agglomeration features are enhanced year by year. This paper attempts to summarize general recommendations that are beneficial to urban eco-efficiency and provide scientific support for urban sustainable development and ecological protection.

众所周知，生态效率是评价城市生态文明建设的重要指标，黄河流域是国家生态文明建设的主要战场。基于2007年至2017年黄河流域43个城市的投入，预期产出和意想不到的产出的面板数据，本文利用基于松弛变量的超效率数据包络分析（DEA）来衡量其生态效率，还结合空间自相关等方法探讨了时空变化格局，并通过实证分析了沿江城市生态效率的时空动态演变及其驱动因素。从2007年到2017年，43个城市的整体生态效率呈现先下降后上升的趋势。大多数城市的生态效率水平相对较低，但仍有很大的改善空间。从流域划分的角度来看，上，中，下城市的生态效率空间格局总体上表明中游地区的生态效率高于上下游地区。研究区域近十年来城市生态效率的莫兰一世表明，生态效率首先呈现空间负相关，然后演变为空间正相关，集聚特征逐年增强。本文试图总结一些有益于城市生态效率的一般性建议，并为城市可持续发展和生态保护提供科学支持。