Poor air quality has harmful effects on human health and determination of pollutant levels in the air is important for human health and the environment. The main purposes of this study are analyzing the winter air pollution of Gaziantep province in terms of some parameters (NO2 and SO2) and the discussing the necessary precautions to be taken. In this context, 5 passive samplings were made at each of the 30 points determined between 2015-2020. Statistical analyzes of the measurement results were made using the SPSS 17.0 V. program and it was determined that there were significant relationships between the results. FWhile sigmoid, logistic, gompertz, sınırlaand linear models were developed with SigmaPlot Version 11.0 software, future predictions of pollutant values were made using linear artificial neural network (ANN) and non-linear artificial neural network (NARX-ANN) in MATLAB R2020a program. In the NO2 and SO2 estimations made for 2040, the most reliable results were obtained with the linear model, and the pollutant concentrations were calculated as 50.91 µg/m3 (R2 =0.9999) and 168.48 µg/m3 (R2=0.9762), respectively. Among the models developed with artificial neural networks, NARX-ANN Bayesian regulation algorithm was found to be more suitable and reliable for predictions (R2=0.8785 for NO2, R2=0.9021 for SO2). Although it has a lower R2 than the linear model, the NARX-ANN model is considered to be more meaningful because it also considers meteorological data in forecast calculations. At the end of the study, suggestions were presented for the improvement and sustainability of Gaziantep’s air quality. These are the homogeneous distribution of population density by urban transformation and city planning, the creation of infrastructures such as public transportation that will reduce traffic density, the increase of incentives and grants for alternative clean fuels, and the establishment of a smart city concept integrating information and communication technologies and fixed/mobile measurement sensors.