The results show that although bacterial DNA profiling may be useful for forensic soil analysis, certain variables, particularly time frame for bacterial collection, must be considered. This report identifies several factors that influence the utility of using TRFLP profiling to link a questioned soil to a crime scene. Profile similarity must necessarily be much higher within a habitat than among habitats compared. Further, habitats cannot be highly heterogeneous over small spatial distances. Also, similarity cannot change substantially over time, since reference sample collection will always occur at some point after the time when the crime occurred. Overlap in similarity indexes among and within habitats will also make use of TRFLP profiling for forensic soil analyses difficult or impossible. In the current study, habitats were on average distinguishable from each other during most of the year; however, the higher among-habitat averages during April and the fall months overlapped within habitat averages for the marsh, woodlot, and sandy woodlot. Further, during several months in the agricultural field, similarity indexes were as low or lower as many among-habitat results. These types of problems could potentially be remedied by analyzing multiple reference samples, as is commonly done in forensic hair analysis, since two soils from the same source (habitat) can be markedly different, but they are unlikely to be highly similar by chance. 4 tables, 3 figures, and 31 references