Air dispersion modeling over coastal regions has proven to be a remarkable challenge in the field of air quality. Many conventional plume dispersion models, such as ISC2 and HYSPLIT, are unable to model such dispersion with the precision that is necessary to accurately predict ground-level concentrations in coastal areas. Considering this, the present work was carried out with two primary objectives: i) to evaluate the effectiveness of currently available mathematical models in predicting plume dispersion over a coastal region and ii) to study the impact of sulfur dioxide emissions from a petroleum refinery over a different community located in the adjacent area. This study demonstrates that CALPUFF predictions are more reliable compared to those of the other models studied, however the operation of CALPUFF is highly data intensive and in many instances, it is difficult to obtain all required input data. This is a particular problem for regions outside ofthe United States of America where sufficient data is difficult to obtain. In addition, the study concluded that the predicted annual average SO2 concentrations in the nearby communities are well within regulatory limits.