The aim of this research was to develop pure predictive models to provide short-term prediction of near surface ozone concentration for the Chennai capital city of Tamilnadu. The short-term prediction of near surface ozone levels is very important due to the negative impacts of ozone on human health, climate and vegetation. A new method for short-term prediction is presented using the neural network technique. Due to increase in industrial and anthropogenic activity, air pollution is a serious subject of concern today. Ground level ozone prediction using the technique of adaptive pattern recognition is developed. The model can predict the mean surface ozone based on the parameters like wind speed, temperature and % Relative Humidity. The Mean absolute Percentage of error of the data during testing is 8.647%. The model can perform well both in training and independent periods.