Control systems under normal conditions can provide a desirable performance. But when faults occur in the system, maintaining the appropriate operating condition is a difficult and often necessary matter. In fact, lack of timely fault detection in sensitive systems will lead to damage significant amounts of resources and information. As a result, a growing tendency in the field of fault detection in both scientific and industrial communities has been created. However, if the system under consideration is nonlinear, fault detection cannot be possible with linear methods. In this case the main difficulty is in accurately modeling of process which effects on the accuracy of fault detection and troubleshooting. Fuzzy systems theory, is an effective tool to deal with the complicated and uncertain situations. This paper has considered the problem of fault detection based on the modeling for nonlinear systems using interval type-2 fuzzy system. Our proposed method for fault detection is to create a confidence bound using the estimation of upper and lower bounds for the system output which can be done using a type-2 fuzzy system. Here, a residual signal is produced which determines the presence or absence of fault in the system. In this method in case of deviating the output graph of the control system from the estimated upper and lower bounds, the occurrence of fault can be detected. Finally, in order to show the capabilities of proposed method, the method is applied on three-tank and electro-hydraulic nonlinear systems and the results are very satisfying.