Frequent occurrences of episodes as a result of rapid urbanization of the cities have become a cause of concern among the air pollution scientists and urban planners. Therefore, a systematic framework for episodic urban air quality management (e-UAQM") is needed to control pollution levels during adverse dispersion conditions. Mathematical modelling techniques are mostly used in predicting the air pollution dispersion phenomena. However, there exist considerable difficulties in describing pollutant dilution mechanism in urban environment due to its complexity and nonlinearity. Majority of the existing conventional models fail in accurately predicting the air pollutant concentration due to inherent limitations in form of various assumptions. In this paper, the application of conventional deterministic, stochastic, hybrid of deterministic and statistical, and artificial neural network (ANN) based line source models in the management of local air quality of Delhi city is explained. The limitations encountered in modelling of urban air pollution are also presented. Copyright © IICAI 2005.