In order to begin to elucidate the genomic basis of the attenuation of mouse-adapted, poliovirus type 2 strain W-2 (PV2/W-2), we have cloned and sequenced the virus and compared it with the virulent, mouse-adapted PV2/Lansing strain. In addition, we have performed computer-generated comparisons of PV2/W-2 to the non-mouse-adapted, attenuated PV2/Sabin strain to determine whether mutational patterns occur that result in attenuation. The PV2/W-2 genome is 7434 nucleotides in length, which is three bases shorter than PV2/Lansing. The 5' non-coding region of PV2/W-2 is 747 nucleotides in length (compared to 744 in PV2/Lansing) and shares 98.8% identity with PV2/Lansing and 82-3% identity with PV2/Sabin. Overall, the PV2/W-2 polyprotein (2205 amino acids) is two amino acids shorter than that of either PV2/Lansing or PV2/Sabin (2207 amino acids). These contiguous deletions fall within the P3-D region (polymerase). Within these 2205 amino acid residues only 26 differences were observed between PV2/W-2 and PV2/Lansing (98-8% identity), whereas 92 occurred between PV2/W-2 and PV2/Sabin (95-8% identity). The 3' non-coding region of PV2/W-2 is 72 nucleotides in length and shares 100% identity with PV2/Lansing and 98.6% identity with PV2/Sabin. Amino acid changes in the capsid protein region occurred in neutralization sites 1 and 3, areas previously shown to be important for pathogenicity. The cleavage site between non-structural proteins P2-C/P3-A consisted of a glutamine-serine pair, in contrast to other sequenced polioviruses which have a glutamineglycine dipeptide.