A full-length reverse-transcribed, infectious cDNA copy of coxsackievirus B3 (CVB3) was used to determine the nucleotide sequence of this cardiotropic enterovirus. Comparison of the nucleotide sequence and the deduced amino acid sequence of the viral precursor polyprotein with the sequences of other group B coxsackieviruses (CVB1 and CVB4) demonstrates a high degree of genetic identity. They share about 80% homology at the nucleotide level and about 90% when the amino acid sequences of the polyproteins are compared. The potential processing sites of the coxsackievirus polyproteins, as deduced from alignment with the poliovirus sequence, are conserved among these enteroviruses with the exception of the cleavage sites between VP1 and 2Apro and between polypeptides 2B and 2C. Comparison of the 5' termini of the enteroviral genomes reveals a high degree of identity, including the initial 5' consensus UUAAAACAGC, suggesting essential functions in virus replication. An important finding concerning the molecular basis of infectivity was that both recombinant CVB3 cDNA and in vitro-synthesized CVB3 RNA transcripts are infectious, although two initial 5' uridine residues found on the authentic CVB3 RNA were missing. Here, we report that cDNA-generated CVB3, as well as CVB3 generated by in vitro-synthesized RNA transcripts, regains the authentic initial 5' uridine residues during replication in transfected cells, indicating that the picornaviral primer molecule VPg-pUpU may be uridylylated in a template-independent fashion. The generation of virus or virus mutants with infectious recombinant CVB3 cDNA and in vitro-synthesized infectious CVB3 transcripts should provide a valuable means for studying the molecular basis of the pathogenicity of this cardiotropic enterovirus.