Faults in the intake and exhaust path of turbocharged common-rail Diesel engines can lead to an increase of emissions and performance losses. Standard fault detection strategies based on plausibility checks and trend checking of sensor data are not able to detect and isolate all faults appearing in the intake and exhaust path without employing additional sensors. By applying model based methods a limited sensor configuration can be used for fault detection. Therefore a model based fault diagnosis concept with parity equations is considered, . In this contribution the fault diagnosis system, which comprises semi-physical thermodynamic turbocharger model, models of gas pressure in the intake and exhaust manifold, residual generation, residual to symptom transformation and fault diagnosis is presented.The residuals are calculated from the difference between the virtual sensors and the measured values and from the difference between the virtual sensors and outputs of reference models which represent the turbocharger quantities in the fault-free operation, , . The created residuals are applied for the detection of leakages, restrictions and clogged actuators in the intake and exhaust path of the Diesel engine. The fault detection is based on deviations of the residuals if they transgress the operation point dependent thresholds. Further the engine operation area is divided into different regions with individual pattern of the residual deviations. The considered faults are isolated by combination of the residual deviations in different regions. Fault diagnosis system is implemented as a set of operation region dependent fuzzy systems suitability for onboard application due to the high level of interpretability and its manageable structure complexity.The model based fault diagnosis is verified on a dynamic engine testbench with 110kW Opel CR-Diesel engine with VGT turbocharger and HPEGR. Leakages and restrictions are implemented in the intake and exhaust path of the engine. Measurement data in faulty and fault-free operation is used for verification of the fault detection. Summing up it is shown with real measurements from the engine testbench that all considered faults can be isolated. The manageable complexity of the introduced models and the diagnosis system shows suitability for onboard application.