The increasing demands on quality of power, emissions and overall performance of combustion engines lay new goals for the hardware and the software development of control systems. High-performance embedded controllers open the possibilities for application of numerical methods to solve the problems of modeling and control of combustion engines. Algorithms for estimation of state and parameters are essential components of many advanced control, monitoring and signal processing engine applications. A widely applicable estimator is given by the Extended Kalman Filter (EKF) which defines a finite memory recursive algorithm suited for real-time implementation, where only the last measurement is used to update the state estimate, based on the past history being approximately summarized by the a priori estimate of the state and the error covariance matrix estimate. The proposed EKF uses an augmented air-path state-space model to estimate unmeasurable mass flow quantities. The EKF algorithm based on the augmented state-space model considerably reduces the modeling errors compared to the open loop estimator simulation and compared to the EKF without the augmentation which is demonstrated on a standard production Diesel engine data. The experimental validation of the observed state quantities is performed against the measured pressures and turbocharger speed data. The estimated mass flow quantities are indirectly validated through the air compressor flow that is directly validated against air mass flow sensor data. Two two-sensor setups are considered in this study. In the first experiment the intake manifold pressure and the turbocharger speed is used. The second experiment uses the intake manifold pressure and the exhaust manifold pressure as the measurement information for the EKF. The second experiment gives more precise mass flow estimate in term of less bias on the estimates, but more variance due to the high frequency exhaust manifold pressure variations caused by the exhaust valves. © 2012 Springer London.