Genes encoding SHV-1 and SHV-2 were sequenced by different methods. Nucleotide sequencing of the coding strand by standard dideoxy-chain termination methods resulted in errors in the interpretation of the nucleotide sequence and the derived amino acid sequence in two main regions which corresponded to nucleotide and amino acid changes that had been reported previously. The automated thermal cycling method was clearly superior and consistently resulted in the correct sequences for these genes.