Innovation is the primary force to drive urban development. In the comprehensive science and technology policy system, talent policy is an important factor affecting urban innovation, as various types of talents provide intellectual support for innovation. Practically, different cities in China have issued a wide range of science and technology talent policies. Against this background, it is of great significance to study the causal paths through which these policies affect urban innovation. However, the existing literature does not provide an answer to this question. On one hand, most studies on science and technology talent policy evaluated the policy itself, few have focused on how such policies exert consequences quantitatively; on the other hand, studies on urban innovation either focused on its measurement, or considered it as a dependent variable to explore the impacts of both government policies and other conditional variables.Based on the related literature, this paper deconstructs science and technology talent policy into ten dimensions, namely, policy maneuverability, talent cultivation, talent flow, talents incentive, talent investment, talent management and service, the guarantee for talents, science and technology innovation culture and environment construction, talent entrepreneurship and talent evaluation. Through collecting and quantifying the policy documents of 27 provincial capitals and 4 municipalities and measuring the urban innovation level with urban innovation index, the crisp set qualitative comparative analysis (csQCA) method is used to analyze the paths through which science and technology talent policy affects urban innovation.It finds that the paths through which science and technology talent policy promotes urban innovation can be summarized into three types: innovation and entrepreneurship-oriented type, cycle-driven type, and regulation and guarantee type. For the first type, city governments pay attention to guide the innovation and entrepreneurship of science and technology talents, to improve the social environment, and to strengthen the construction of urban infrastructure related to the diffusion of knowledge and technology. For the second type, out of the concern for science and technology talents themselves, city governments emphasize the cultivation, motivation, and service for talents, and by promoting the transformation of science and technology achievements to back feed the resource input, to continuously enhance the level of urban innovation. For the third type, city governments emphasize to maintain order by regulating the market, and to guarantee the life of talents.In terms of policy implications, it is completely different to improve the degree of marketization from improving the market normative level. The former is to put science and technology talents into the market environment for competition. The fiercer the competition is, the more obvious the trend of fittest survival will be. However, it may violate fairness and endanger market order. The latter is to impose various restrictions on market competition with the policy tool as imposed hand. By regulating the talent market, science and technology talents in various industries and fields can make progress side by side and hence achieve the goal of urban innovation. Since there are huge gaps in technological innovation resources as well as technological development level among different regions and cities in China, local governments should choose corresponding paths to carry out innovation activities, to enhance innovation output, and to improve the level of urban innovation.This paper makes a systematic exploration on the internal mechanism through which science and technology talent policy impacts on urban innovation for the first time. It contributes to the literature in the following aspects: first, it applies the csQCA method and conducts cross case causality analysis of different cities. The urban innovation results acquired under different combinations of science and technology talent policy elements are helpful to theoretically enrich the literature on both science and technology policy and urban innovation. Second, depending on the specific situation and configuration, it finds three types of influence paths. Conclusions from this research can provide references for local governments to design suitable science and technology talent policy and hence promote urban innovation practically.