Internal combustion engines control systems are evolving rapidly in order to meet the most recent emissions standards: this process requires a deep knowledge of how the combustion process takes place, since heat-release control is crucial to manage the trade-off between engine-out emissions and best suit the tailpipe aftertreatment system operating point.Nitrogen oxides (NOx) have currently become a critical pollutant emission that needs to be limited in compression-ignited engines. Since a selective catalytic reduction (SCR) system is present in several applications, engine-out NOx concentration is a fundamental parameter to be evaluated.This work shows how an estimation of NOx concentration can be deduced from instantaneous in-cylinder pressure measurement and some of the parameters, related to the intake charge, currently available on electronic control units (ECU). A prediction model, based on Zeldovich mechanisms and Arrhenius rate of combustion is proposed, which exploits as main contributions: rate of heat release and adiabatic flame temperature. An experimental campaign (DOE) has been carried out on a diesel engine, varying the main engine control parameters, to tune the model in steady-state operating point. The predictive capability is then evaluated by feeding the model with a validation dataset in both steady state and transient condition. Finally, model response to measure uncertainties is discussed.