Mean Value Engine Models (MVEM) represent average behaviour of an engine over one or more thermodynamic cycles and have been designed for automotive control and diagnosis applications. However, most MVEMs are limited to the description of the dynamics of few engine sub-systems. The diagnostic capabilities of a vehicular engine health management (VEHM) system that uses such MVEMs are limited. In this paper, the process of deriving an MVEM for an entire engine system from an instantaneous within-cycle crank-angle model (WCCM) is described. This is expected to be more beneficial for fault diagnosis in VEHMs since such MVEMs in the context of state observers, can be used to detect a broader range of faults and also generate a larger number of fault signatures for better fault detection and isolation (FDI). Extended Kalman Filter (EKF) based estimators are developed that use this MVEM for state estimation. The quality of estimation results was significantly improved by modelling the inaccuracy of the MVEM as a bias and extending the MVEM with the knowledge of these bias tables. Fault residue generators are designed using various MVEM-based fault models in a state observer.