The interest for NOx estimators (also known as virtual sensors or inferential sensors) has increased over the recent years due to benefits attributed to cost and performance. NOx estimators are typically installed to improve On-Board Diagnostics (OBD) monitors or to lower bill of material costs by replacing physical NOx sensors. This paper presents initial development results of a virtual engine-out NOx estimator planned for the implementation on an ECM. The presented estimator consists of an airpath observer and a NOx combustion model. The role of the airpath observer is to provide input values for the NOx combustion model such as the states of the gas at the intake and exhaust manifolds. It contains a nonlinear mean-value model of the airpath suitably transformed for an efficient and robust implementation on an ECM. The airpath model uses available sensory information in the vehicle to correct predictions of the gas states. The NOx combustion model is a crank-angle resolved model of the incylinder processes, consisting of a pressure-heat release model, zone temperature model and NOx formation model. The NOx combustion model operates in open-loop mode and it is calibrated in offline mode using instrumentation grade in-cylinder pressure sensor. The presented work includes detailed description of the model, explanations of design of experiment, calibration procedure and available validation results.