During replication of polyoma DNA in isolated nuclei, RNA was found attached to the 5' ends of growing progeny strands. This RNA starts with either ATP or GTP and can be labeled at its 5' end with (32)P from beta-labeled nucleotides. Digestion of progeny strands with pancreatic DNase released (32)P-labeled RNA that, on gel electrophoresis, gave a distinct peak in the position expected for a decanucleotide. We believe that this short RNA is involved in the initiation of the discontinuous synthesis of DNA and propose the name "initiator RNA" for it. The covalent linkage of initiator RNA to 5' ends of growing DNA chains was substantiated by the finding that (32)P was transferred to ribonucleotides by alkaline hydrolysis of purified initiator RNA obtained by DNase digestion of polyoma progeny strands synthesized from [alpha-(32)P]dTTP. While initiator RNA was quite homogeneous in size, it had no unique base sequence since digestion with pancreatic RNase of initiator RNA labeled at its 5' end with (32)P released a variety of different [(32)P]oligonucleotides. The switch from RNA to DNA synthesis during strand elongation may thus depend on the size of initiator RNA rather than on a specific base sequence.