As part of a testing exercise performed in conjunction with a national conference of the Canine Accelerant Detection Association, 34 dogs were subjected to a series of tests. In one test, the dogs were nearly unanimously successful in locating one of five cans containing 50 percent evaporated gasoline on a burnt carpet matrix and in pinpointing a 6-inch-square sector on a piece of plain carpeting where the same amount of gasoline was applied. Only half of the drugs, however, were able to detect a second doped sample containing a lesser amount of gasoline and they registered a number of alerts on samples containing only burnt carpeting material. The dogs were also tested on measured amounts of other light, medium, and heavy petroleum products applied to a variety of substances containing significant pyrolysis products. As a group, the dogs were much less successful in pinpointing these products than they were with gasoline at the same level and again registered a number of alerts on cans containing only pyrolysis products. The significant number of alerts by the dogs on samples not containing gasoline or other products demonstrated the importance of obtaining laboratory confirmation of samples on which dogs alerted and the need to keep accurate field and training records of the dogs to establish their credibility. 11 references, 5 tables, and 1 figure