本文以福建省九地级市科技创新系统为研究对象，运用超效率数据包络分析视窗模型分析2010-2014年各城市科技创新效率的动态变化，联合Malmquist指数揭示城市科技创新效率变动的深层次原因。研究表明：福建省多数城市科技创新效率DEA无效，效率提升空间较大。福州、厦门等经济较发达地区的科技创新效率相对较低，福州投入拥挤效应比较显著。Malmquist指数分析表明，技术进步始终是福建各地级市全要素生产率增长的制约因素，其综合技术效率的提高更多地得益于规模效率的改进。据此，提出提升福建科技创新效率的若干建议。 

 Science and technology (S&T) innovation system of 9 prefecture-level cities in Fujian Province is taken as a research object in this paper. Super efficiency-DEA window model is applied to analyze the dynamic changes of various cities’ S&T innovation efficiency from 2010 to 2014. The fundamental causes of these changes are explored with Malmquist index. The results show that S&T innovation efficiencies for most of the cities are DEA invalid, and the improvement space is large. Innovation efficiency of economically-developed cities such as Fuzhou and Xiamen is relative low. There is a significantly crowding effect of investment in Fuzhou. The index analysis shows that technological progress is always a key restriction factor in the growth of total factor productivity, and comprehensive technical efficiency improvement is more determined by scale efficiency improvement. Finally, some suggestions are proposed to improve the S&T innovation efficiency of Fujian. 