Terminal restriction fragment length polymorphism (TRFLP) is an increasingly popular method in molecular ecology. However, several key limitations of this method have not been fully examined especially when used to study fungi. We investigated the impact of spore contamination, intracollection ribosomal DNA internal transcribed spacer (ITS) region variation, and conserved restriction enzyme recognition loci on the results produced by TRFLP to characterize soil fungal communities. We find that (i) the potential for nontarget structures such as spores to contribute DNA to target sample extractions is high; (ii) multiple fragments (i.e. 'extra peaks') per PCR primer-restriction enzyme combination can be detected that are caused by restriction enzyme inefficiency and intracollection ribosomal DNA ITS variation; and (iii) restriction enzyme digestion in conserved vs. variable gene regions leads to different characterizations of community diversity. Based on these results, we suggest that studies employing TRFLP need to include information from known, identified fungi from sites within which studies take place and not to rely only on TRFLP profiles as a short cut to fungal community description.