Electrochemical gas sensors with the ability to detect the concentration of individual gases in a multi-gas environment are in huge demand in many areas of technology, such as threat detection, industrial quality control and, importantly, vehicular emission monitoring. While oxygen l sensors are used for on-board diagnostics in spark-ignition vehicle emission systems, a similar sensor is not available for NOx, NH3 and non-methane hydrocarbon (NMHC) gases found in diesel engine exhaust. LANL has developed patented pre-commercial prototype mixed-potential sensors with strong sensitivity and selectivity towards these target gases, but a single device with absolute selectivity is still elusive (1, 2). To address this pressing problem, we have tested and successfully demonstrated the utility of employing a sensor array rather than individual sensors, in conjunction with a first-principles data analysis approach. Our computational framework was employed to predict the composition of gas mixtures containing two individual gases, both of which affect sensor readings, with less than 2% average error (3). Here we present the results of our model trained on three and four-gas mixtures involving NO, NO2, NH3 and C3H6. The three sensors used in the array comprise a dense La0.8Sr0.2CrO3 or AuPd sensing electrode and dense Pt electrode with porous YSZ electrolyte. A physically motivated model is used to analyze the data and find the composition of the gas mixture. This work aims to provide robust and accurate estimates of NOx, NH3 and unburnt non-methane hydrocarbon content in diesel vehicle exhaust, laying a foundation for on board, real-time sensing applications. Acknowledgments The research was funded by the Los Alamos National Laboratory Directed Research Development Exploratory Research (LDRD-ER) program. References R. Mukundan, E. L. Brosha, F. H. Garzon, US 7,575,709 (2009).R. Mukundan, E. L. Brosha, F. H. Garzon, Journal of The Electrochemical Society 150, H279 (2003).K. P. Ramaiyan, C. R. Kreller, E. L. Brosha, R. Mukundan, U. Javed, A. V. Morozov, ECS Transactions (2016) 75, 107-111. LA-UR-17-23016