Abstract In order to assess the impact on air quality due to the burning oil wells and to develop an early warning system by forecasting the movement and dispersion of the plumes, the Air Resources Laboratories Atmospheric Transport and Dispersion (ARL-ATAD), a regional transport model, was used. Based on the information on the number of wells on fire and their emission factors, an emission inventory of soot particles was developed and their values were periodically adjusted. It was observed that the region between south Kuwait and Safaniyah was most affected by the smoke trajectories during March May 1991. Monthly average simulated pollutant concentration and deposition contours showed that the concentration pattern for the region changed with the changes in climatic conditions and the source emissions. Using real and forecasted synoptic and upper air meteorological data, trajectories were produced at 980, 850 and 700 mb pressure levels. Heavy and light smoke concentration zones were identified and air pollution forecast maps were prepared. In order to compare the results of the air pollution forecasting system with the real concentration pattern of the smoke, satellite images of the smoke were obtained. By comparing the concentration contours with the satellite images, it was observed that the model results follow more or less the same pattern as the real smoke plume movement. The daily average observed PM10 values were also compared with the simulated smoke concentration pattern. The forecast turned out to be reliable about 80°o of the time and the performance of the model proved to be quite satisfactory.