A robust fault detection and isolation (FDI) approach for uncertain constrained nonlinear systems (CNS) is outlined. The FDI scheme is based on a second order sliding mode observer (SMO). First, the closed-loop system is formed by incorporating the constraint terms into the system’s dynamic equation and then the observer is designed based on the resulting closed loop system. Stability of the SMO resulting from the utilization of equivalent control concept is then proved by assuming that the considered uncertain CNS has a single output and two outputs, respectively. An attractive feature of the FDI methodology is that the diagnostic observer can directly supply the estimate of the faults, hence, making the fault isolation a simple task. Finally, an example is given to show the effectiveness of the proposed SMO based FDI strategy.