Several models have been proposed to analyze and predict levels of pollutants in the ambient air.The artificial neural network (ANN) is one of these methods. ANN, which is a branch of artificial intelligence, has spread among researchers both because of its nonlinear mathematical structure and its ability to make accurate predictions. We compared the abilities of two different types of ANN, the multilayer perceptron (MLP) and the radial basis function (RBF), to forecast particulate matter (PM) with diameters of 2.5 microns or less (PM 2.5 ) based on meteorological data from District 20 of the municipality of Tehran (Shahre Ray City). The input data were hourly air temperature, wind speed, and percent humidity, and output was PM 2.5 concentrations. The mean bias error (MBE) of the results were 0.0503 and 0.0032 in the MLP and RBF networks, respectively. The coefficient of determination (R 2 ) and the index of agreement (IA) between the observed data and the predicted data were 0.954 and 0.987, respectively, for the MLP, whereas for the RBF the R 2 was 0.99 and the IA was 0.998. Sensitivity analysis performed for the MLP indicated that percent humidity is the most important factor in the prediction of PM 2.5 . (1995) predicted the ozone (O 3 ) levels at five stations in Mexico City using a neural network and nonlinear regression. Tasadduq, Rehman, and Bubshait (2002) applied an MLP neural network to predict the hourly average temperature in Jedah, Saudi Arabia. The results of their study indicated the suitability of their learning algorithm, which was "backpropagation," or "backward propagation of errors." Zickus, Greig, and Niranjan (2002) used four machine learning methods, including logistic regression, ANN, multivariate regression, and decision trees, to predict daily concentrations of PM 10 in the air in Finland. Zickus et al.'s results indicated that for predicting air pollution levels, the first three methods provided favorable performances compared to the decision tree method. Kukkonen et al. (2003) applied an ANN, PM, along with other air pollutants, pose serious hazards to human health.