As a means of improving the efficiency of law enforcement and accountability, the American police big data promotes the transformation of police practice from traditional law enforcement activities to modern intelligence activities, but to what extent it improves the effectiveness of police practice is still an open empirical issue. This paper discusses the historical evolution of data-driven police practice, and introduces the practical progress of big data in American police from the aspects of its rising background, data characteristics, technical tools and practical applications. The paper examines four risk issues in the practice of American police big data, and then expounds the policy enlightenment from three aspects: data quality control, data safety management and data analysis technology.