Abstract: County-scale eco-efficiency has been recognized as one of the important base unit that supports regional green development. Taking the urban agglomeration in the middle reaches of Yangtze River as the study area, this paper constructed the input-output index system of eco-efficiency based on multi-source data, used the Super-EBM model to measure the county eco-efficiency from 2005 to 2018, and analyzed the dynamic characteristics of its spatial pattern with the help of LISA time path and LISA temporal and spatial transition methods. The spatiotemporal transition types were nested with the panel quantile regression model to explore the driving mechanism. The results showed that the overall level of county-scale eco-efficiency of the urban agglomeration in the Middle Reaches of the Yangtze River increased significantly during 2005 to 2018. The agglomeration development effect of the three provincial capitals as the center was significant. Urban agglomeration in the middle reach of county-scale eco-efficiency of spatial distribution of cohesion was stronger and had a higher path lock feature, there was a transition in each county unit inertia, the local spatial characteristics of eco-efficiency are more dynamic in connecting county units of “Wuhan Metropolitan Circle-Changzhou-Zhuzhou-Xiangtan Urban agglomeration”, Fengcheng in space depended on the direction of maximum volatility, Changsha municipal districts had maximum stability. Four driving-restraining models, including economy-climate drive, industry-resource control, economy-resource control, and industry-precipitation control, presented an evolution pattern from “Wuhan City Circle-Poyang Lake Urban Agglomeration” to “Changzhou-Zhuzhou-Xiangtan Urban Agglomeration” from restriction to drive, each containing different driving mechanisms.