A leaf area index simulation model of cotton was developed under full and deficit irrigation in Xinjiang. The model was based on beta function of daily heat effect as time scales, and the effect of soil water stress was considered. One ordered derivative of the Logistic function was used to describe the changing rate of leaf area index. One ordered derivative of the Logistic function was also used to describe leaf senescence process and assumed that leaf senescence occurred since the early flowering. The acceleration effects of soil water stress and temperature on leaf senescence were also considered in the model. Finally field observations from Shihezi were used for model parameter calibration, evaluation and sensitivity analysis. Evaluation results showed that under the condition of full irrigation the root mean square error (RMSE) between simulated and observed leaf area index (LAI) values was 0.22m2/m2 and the residual accumulation coefficient (CRM) was -0.01. Under the condition of 20% and 40% deficit irrigation RMSE and CRM were 0.37m2/m2, 0.05 and 0.05m2/m2, 0, respectively. In addition, the determination coefficient of linear relationship between simulated and observed LAI was 0.96. Simulation results showed that the model was accurate in describing the dynamic changing process of cotton LAI during the whole growth period in Xinjiang. Sensitivity analysis results showed that under the condition of sufficient water, potential leaf aging area, maximum daily leaf extension area and aging area, early flowering start time were the main parameters affecting the model. The model had a few parameters and it can be used to supply the LAI dynamic information for establishing the Xinjiang cotton growth model and evapotranspiration model.