Emissions regulations for heavy-duty diesel units used in maritime and power generation applications have become very strict the last years. Hence, the industry is enforced to limit specific gaseous and particulate emissions (NOx, SOx, COx, PM and HC) depending on the regulations. Among numerous methods, simulation models are extensively used to support the development of techniques used for the control of emitted pollutants. This is very important for large-scale engines due to the extremely high cost of the experimental investigation resulting from the size of the engines and the test equipment involved. Beyond this, simulation models can also be used to support NOx monitoring, since on-board verification techniques are to become mandatory for the marine industry in the near future. Last but not least, simulation models can also be used for model-based control applications to support the operation of both in-cylinder and after-treatment techniques. Currently, the major controlled pollutant for both marine and stationary applications is NOx. For this reason, in the present work, authors focus on the development and application of a simplified NOx, model with special emphasis on its ability to predict the effect of operating conditions on NOx, for both two and four-stroke diesel engines. To accomplish this, an existing well validated simplified NOx, model has been modified to enhance its physical background and applied on 16 different large-scale diesel engines utilizing 18 different sets of measurements acquired at field. The engines considered were marine propulsion engines, marine auxiliary engines and stationary engines used for power generation on the Greek Islands. The proposed model is a simplified, physically-based, semi-empirical, pseudo-multi-zone one which makes use of the engine's measured cylinder pressure trace. Using this, it estimates the fuel combustion rate and the resulting fuel amounts are attributed to combustion zones which are burnt separately. The amount of air in each zone is determined from the local fuel-air equivalence ratio using an empirical correlation developed in the present work. Each zone, after its generation, behaves as a closed thermodynamic system and continues to expand inside the combustion chamber contributing to NOx, formation which is estimated using the extended Zeldovich mechanism. The total amount of NOx, emissions at the engine exhaust is estimated from the integral of zone NOx, at exhaust valve opening. Additionally, the model provides the time history of NOx, formation inside the combustion chamber. As revealed form the specific application, the proposed model captures qualitatively and after calibration quantitatively NOx, variation with engine operating conditions and start of injection, which is very important for large-scale diesel engines. 75% and 98% of the test cases examined present a relative error less than 10% and 30% respectively. This encourages model's use for engine development and real-time NOx prediction. (C) 2014 Elsevier Ltd. All rights reserved.