: This paper presents efficient methods for online fault detection and diagnosis of Network-on-Chip (NoC) switches. The fault model considered in this research is a system level fault model based on the generic properties of NoC switch functionality. The proposed method is evaluated by fault simulation in a platform using this system level fault model. The experimental results show that with a relatively low area overhead, a large number of NoC switch faults can be detected and diagnosed. Key-Words: Network-on-Chip, Switch, Online Testing, Functional Fault Model, Fault Diagnosis. Categories: B.4.3, B.4.5, B.8.1