This paper proposes a method to detect an intake manifold leakage for a Diesel engine with a dual loop EGR system. The intake manifold leak has a strong impact on the engine performances by changing the intake manifold burned gas ratio. This fault is analyzed according to the control structure used and also according to the EGR operating mode. The paper proposes a diagnosis algorithm to detect the intake manifold leak in sequential or simultaneous use of the two EGR paths. The sensors considered are the mass air flow meter, the intake manifold pressure sensor, the exhaust equivalence ratio sensor and the differential pressure sensor (across the HP EGR valve). The diagnosis is based on a criteria that uses the redundancy between these sensors and air system models or estimators. The diagnosis threshold depends on the engine operating conditions as well as the sensor or model dispersions. The proposed algorithms is implemented online using an approximation of the diagnosis threshold computed with Monte Carlo simulations. The paper ends with results carried out from experimental and simulated data of European drive cycle. The results demonstrate that a leakage detection is effective with the considered set of sensor. It is also possible to define the engine operating conditions where the diagnosis is achievable.