To determine whether a fire has been intentionally set, it is important for scene investigators to identify accelerants recovered from fire debris. However, highly volatile accelerants are not effectively recovered by sampling methods used to recover debris at a fire scene because these accelerants diffuse in the air and do not remain in the fire debris. This study tested the effectiveness of identifying highly volatile accelerants by sampling air at a fire scene using a portable gas pump. This method is compared to sampling with the solid phase microextraction method (SPME) and dynamic headspace sampling in order to compare the sensitivity between air sampling and sampling from fire debris. Air sampling was performed by retaining accelerants in stainless steal tubes using stainless steal gauzes and a gauze retaining spring. The tubes were filled with Tenax TA and Carbopack B. The results indicate that when highly volatile compounds are used as accelerants, the chromatogram obtained from fire debris is complicated. As such, the use of air sampling with a portable gas pump was found to be more effective than the SPME or dynamic headspace sampling of fire debris. Figures, tables, references