The preS2/S genes of hepatitis B virus isolated from 29 acutely or chronically infected individuals in the Gauteng province of South Africa were sequenced. Phylogenetic analysis of these sequences in comparison with global isolates from the GenBank database showed that 24 sequences clustered with genotypic group A, three with genotypic group D and one each with genotypic groups B and C. Group A isolates had greater identity with groups D (variation of 6.6%) and E (6.8%) than with the Eastern groups B (7.4%) and C (8.1%) and were most different from group F (11.0%). Of the South African group A specimens, 59.1% clustered with two global sequences to form a discrete segment which we have called subgroup A. The amino acid differences that set these isolates apart from the rest of group A tended to cluster in the preS2 region (amino acids 7, 10, 32, 35, 47, 48, 53 and 54), with a few changes occurring in the major surface antigen (amino acid sites 207 and 209). Analysis of isolates showed that there was a 9-fold higher prevalence of the ay determinant in South Africa than previously reported.