Several mesoscale burns were conducted in 1994 in Mobile Bay, AL, to study various aspects of diesel fuel burning in situ. The target PAHs in the diesel, residue, and soot samples collected during each burn were quantitatively characterized by GC/MS. A simple model based on mass balance of individual petroleum PAHs pre-and postburn was proposed to estimate the destruction efficiencies of the total petroleum PAHs. This study demonstrates the following: (1) Distributions of PAHs in the original diesel and soot were very different. (2) The average destruction efficiencies for the total target diesel PAHs including five alkylated PAH series and other EPA priority unsubstituted PAHs were greater than 99%. (3) Using the model, 27.3 kg of the diesel PAHs were destroyed for each 1000 kg of diesel burned. These were mostly two-and three-ring PAHs and their alkylated homologues. (4) Combustion also generated trace amounts of high molecular weight fiveand six-ring PAHs as well as the four-ring benz[a]anthracene. But the total mass of these pyrogenic PAHs was found to be extremely low: only 0.016, 0.032, and 0.048 kg of the five-and six-ring PAHs were generated by combustion in the three different scenarios for each 1000 kg of diesel burned. From these points, we conclude that in situ burning is an effective measure to minimize the impact of an oil spill on the environment, greatly reducing exposure of ecosystems to the PAHs of spilled oils. A new "pyrogenic index", Σ(other three-to six-ring PAHs)/Σ(five alkylated PAHs), is proposed (see the text for the definition) as a quantitative indicator for identification of pyrogenic PAHs and for differentiating pyrogenic and petrogenic PAHs. Also, this index is demonstrated to be a useful tool for distinguishing heavy fuels from crude oils and light refined products. This method, combined with other criteria, is expected to be applicable to such situations as oil spill investigations, site assessment, and apportioning of legal responsibility for pollution cleanup.