The air quality problems over Ystanbul are related to the low-quality fossil fuel consumption and atmospheric conditions. Sulfur dioxide concentration levels are investigated over Ystanbul to assess air pollution during the heating seasons in which the concentration of air pollutants reach high levels due to the consumption of low-quality fossil fuels. Results reveal that the consumption ratio of coal/fuel-oil has increased drastically in 1980s, Optimum interpolation technique, kriging, is used to obtain the spatial distribution of sulfur dioxide over the area. The resultant sulfur dioxide concentration fields showed three critical regions; Halic basin and Pipli-Taksim area on the European side and Goztepe on the Asian side. It is found that there is a considerable decrease in air pollution levels over Ystanbul in the 1995-1996 heating season. Important factors that have been responsible in this decrease of pollutant levels are found to be the favorable weather conditions, switching to natural gas in many buildings and the consumption of pre-treated coal. For the prediction of air pollution, neural networks are used as the modeling tool and the results are found to be very encouraging. A data set of one month period is used which contains meteorological and air pollution parameters. An average error between the actual and the predicted sulfur dioxide concentration levels of 12% shows us the reliability of the neural network modeling.