Using China’s provincial data from 1991 to 2005, this paper analyzes the impact of urban income disparity on their consumption based on static and dynamic panel data models and state space model. The GMM and Kalman Filter methods are used in the estimation and the variables such as income and price are controlled. The empirical results show that the elasticity of permanent income to consumption is much higher than that of temporary income; and the impact of income disparity on consumption is negative and substantial. A rise of 0.01 in the absolute value of Gini coefficient will cause a reduction of 0.35% in consumption on average. The effects fluctuate with the change of economic structure, consumption expectation and economic cycle. In the beginning years of 1990s, it is positive to enlarge income disparity moderately for consumption. It is the year of 1996 that the negative effect first appears in China. During 1998–2004, an increase of 0.01 to the absolute value of Gini coefficient will result in the reduction of consumption to fluctuate between 0.37% and 0.54%. In order to enlarge domestic demand and promote consumption, the focus should be the improvement of permanent income instead of temporary income, and the vigorous policies to reduce income disparity.