A TANDEM repeat of the sequence Ser-Pro-Thr-Ser-Pro-Ser-Tyr has been found at the C terminus in the largest subunit of RNA polymerase II (refs 1-5) with, for example, 26 units in yeast and 52 in mammals. By removal of this 'tail', it has been shown that 11-23 units are necessary for the normal functioning of the polymerase. The functional role of the repeat is however, unclear, although it has been proposed that it binds to transcription factors. As discussed in an earlier paper, the repeat unit contains two Ser-Pro sequences which seem to be related to a DNA-binding unit found in histones, Ser-Pro-Lys-Lys, and to the Ser-Pro-X-X motif which is often found in gene regulatory proteins and which, it has been proposed, is also a DNA-binding unit. Here, I show that the repeat does indeed bind DNA and present evidence that it does so by the intercalation of tyrosine residues. These experiments involved synthetic peptides containing one or two repeat units. As the sequence Ser-Pro-X-X (where X represents any amino acid) has a strong tendency to assume a special beta-turn, a model of the unit composed of two such beta-turns was made and compared with the structure of the drug Triostin A which is known to intercalate into DNA. Two tyrosine side chains of the repeat overlap well with two quinoxaline rings of the drug and therefore, the model can provide a good explanation of the experimental results.