For many years the attempts to identify functional elements in genomic sequences through motif over-representation have been problematic, as every procedure to isolate such motifs resulted in a very large number of candidates with highly significant p-values. In this paper we postulate that most of these elements originate in ancient transpositional activity, with copies becoming so broken over time that they cannot be recognized as such any more. We have written a software tool which efficiently locates possible ancient repeats and, while the nature of much of our findings in real genomic sequences have yet to be confirmed by further biological analysis, our test results on well understood data sets have been encouraging. © 2007 IEEE.