Interspersed repeats, mostly resulting from the activity and accumulation of transposable elements, occupy a significant fraction of many eukaryotic genomes. More than half of human genomic sequence consists of known repeats, however a very large part has not yet been associated with neither repetitive structures nor functional units. We have postulated that most of the seemingly unique content of mammalian genomes is also a result of transposon activity, written software to look for weak signals which would help us reconstruct the ancient elements with substantially mutated copies, and integrated it into a system for de novo identification and classification of interspersed repeats. In this manuscript we describe our approach, and report on our methods for building the consensus sequences of these transposons.