Recently, numerous researchers and Original Equipment Manufacturers (OEMs) have developed diesel engine-out nitrogen oxides (NOx) estimation algorithms that are capable of running in real-time on production Electronic Control Units (ECUs). These are generally referred to as virtual sensors or inferential sensors. 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 reviews the literature of on-ECU NOx models in order to document the state of the art and identify directions for future work. The discussion includes applications of NOx estimators, accuracy of NOx estimators, required sensor inputs, sources of error, calibration effort, and ECU resource consumption.