This paper presents a survey on model-based fault diagnostics and prognostics for wheeled mobile robots (WMRs). Numerous cases of faults diagnosis and prognosis solutions show that, for a complex machine like a WMR, model-based methods that utilize analytical models of the system are desirable because analytical models can well represent the nominal behaviours of the system. On the other hand, datadriven methods are not practical for WMRs since they require a lot of failure cases and data that are too costly, if possible at all, to obtain. Fault prognosis for a system in general and for WMRs in particular is still at the preliminary stage. System models can also be used to predict the remaining useful life of the system.