Repeat sequences — mainly transposable elements — have been estimated to comprise ~50% of the human genome. Typical approaches for identifying repeats are based on finding sequence similarity to a single consensus sequence for each repeat type. These authors re-examined the human genome using an algorithm that instead relies on relatedness within entire groups of evolutionarily diverged repeats. This led them to an increased estimate of 66–69% for the proportion of repeat-derived sequence in the human genome, after correcting for false positives. These results imply that repetitive DNA may have played a larger part in human evolution than was previously assumed.