Abstract: In the era of knowledge economy, innovation is key for countries and cities to pursue high-quality development and enhance international competitiveness. Atypical knowledge combination, which is novel, exploratory and forward-looking, is considered to be one of the constituent elements of breakthrough innovation and is gradually paid increasing attention by geographers. Under the background of high-quality development in China, the identification of atypical knowledge combination and understanding its driving factors have great practical significance for cities to formulate innovation strategies, improve innovation level and decision-making efficiency. By using the panel regression model of patent data from 2000 to 20018, this research investigates the effects of technology related variety, technology unrelated variety and external innovation cooperation on the output of atypical knowledge combination in the Yangtze River Delta region. The main analysis is based on city-level patent application data issued by China National Intellectual Property Administration and the Z-value method is applied to determine the atypical knowledge combination, and the entropy index first introduced by Castaldi and Frenken is used to measure related variety and unrelated variety. The results show that: (1) The atypical knowledge combinations of Yangtze River Delta region are concentrated in three major technology fields including “Performing Operations & Transporting”, “Chemistry & Metallurgy” and “Human Necessities”, especially in strategic emerging industries. (2) Related variety has a significant positive effect on the output of atypical knowledge combination in cities. In contrast, unrelated variety has no influence on the output of atypical knowledge combination in high hierarchy cities, unrelated variety has an inverted “U” relationship with the output of atypical knowledge combination in general hierarchy cities. (3)There is a significant inverted “U” relationship between external innovation cooperation and the output of atypical knowledge combination in cities. Unrelated variety and external innovation cooperation weaken each other’s positive impact on the output of atypical knowledge combination.