Abstarct: The study many of scientific phenomena is impossible in the laboratory. Hence this phenomenon explains by mathematical model and simulate by using computer code. This model consists of ordinary differential equations or partial differential equations set that solved by using numerical methods this is done. The computer program or code that is able to solve numerically this system of equations, say computer models and run repeatedly this model with different values of inputs say computer experiment. The structure of these models has a special complex that one of its agents, is the large number of inputs. The input variables can be single or interaction with other inputs can affect in outputs of model. Thus, identify the influence of inputs and the inputs are less important or insignificant is necessary. This problem was studied In format as a sensitivity analysis of computer model. various methods have been proposed for sensitivity analysis of models. One method them is sensitivity analysis of variance - baxsed. The sensitivity analysis in this method is done by using of sensitivity measure that famous first-order sensitivity index and sensitivity index total. These indicators are baxsed on the concepts of multiple integrals and conditional variance and conditional expectation is provided. The real value of this integral equation due to the absence of explicit input - output model, it is unknown. The problem arises in the estimation or approximation of integrals. For this purpose, several approaches have been proposed by researchers. Variance baxsed method sensitivity analysis represent tow general approach yclept Monte Carlo baxsed method Saltlelli and Random balanced design for estimated sensitivity index’s. After introducing these two methods, calibration (validation) was performed using the analytic function Sobol. The results showed that these two methods have sufficient accuracy and suitable performance for estimating the sensitivity indices. These methods was used to analyze sensitivity of environmental the ICA-. The computer model ICA- simulated the flow of nitrogen input river water. itrogen is pollution of water resources that an adverse effect on human health and the animals. In this study, the model as a function of the seven input variables (itrogen transformation rates) and a variable output (average annual load of nitrogen entering the river) was considered. Sensitivity analysis of the itrogen Input in Tweed River England, using the Monte Carlo baxsed method of Saltlelli and Random balanced design was showed that four variables, the nitrate uptake by plants, the rate of Denitrification, Immobilization and mineralization were the most important factors affecting on This pollutant the water river teewd .