Terminal Restriction Fragment (TRF) pattern analysis has become a widely used and informative tool for studying microbial communities. Variation between sequence-determined or true TRF length and observed TRF length (TRF drift) has been previously reported and can significantly affect identification of bacterial species using TRF lengths predicted from sequence databases. In this study TRF drift was determined for 21 bacterial species using an ABI 310 Genetic Analyzer. TRF drift was positively correlated with true TRF length and negatively correlated with TRF purine content. This implies that subtle differences in molecular weight, whether from purine content or dye label, can significantly affect the observed TRF length.