This paper attempts to investigate comprehensively, a “U”-shaped relationship between income inequality and crime rates in China after building a cost-benefit analysis model, by using time series data from 1981–2012 and panel data from 1999–2012. The empirical results show that: firstly, in the time series model, the U-shaped relationships between inequality and the total crime rate and rates of various crimes except from smuggling, are very significant in the period of 1981–2012, secondly, the panel threshold models show that inequality and crime tend to be correlated positively with each other during 1999–2012, because the inequality level during this period is much higher than the turning points of inequality estimated in the time series models, although three regions with different development levels are located in different parts of a U-shaped curve between inequality and crime.