Air quality prediction is highly important in view of the health impacts caused by exposure to air pollutants in urban air. This work has presented a model based on support vector machine (SVM) technique to predict daily average carbon monoxide (CO) concentrations in the atmosphere of Tehran. Two types of SVM regression models, i.e. -SVM and -SVM techniques, were used to predict average daily CO concentration as a function of 12 input variables. Then, forward selection (FS) technique was applied to reduce the number of input variables. After converting 12 input variables to 7 using the FS, they were fed to SVM models (FS-(-SVM) and FS-(-SVM)). Finally, a comparison among SVM models operation and previously developed techniques, i.e. classical regression model and artificial intelligent methods such as ANN and adaptive neuro-fuzzy inference system (ANFIS) was carried out. Determination of coefficient (R2) and mean absolute error (MAE) for -SVM (-SVM) were 0.87 (0.40) and 0.87 (0.41), respectively, while they were 0.90 (0.39) and 0.91 (0.35) for ANN and ANFIS, respectively. Results of developed SVM models indicated that both FS-(-SVM) and FS-(-SVM) regression techniques were superior. Furthermore, it was founded that the performance of FS-(-SVM) and FS-(-SVM) models were generally a bit better than the best FS-ANFIS and FS-ANN solutions for short term forecasting of CO concentrations.