Surface (S) gene mutations are clinically important in both hepatitis B VIruS (HBV) prevention and diagnosis. Several HBV vaccination progr~es in endemic countries have revealed their influence on failure of immunisation. Furthermore, sera harboring these variants can escape serological detection, but, remain detectable by HBV-DNA testing. According to these findings, the current work can be divided into two main parts. In the first, the impact of vaccination on the prevalence of S gene variants and, interestingly, how to use HBsAg variability as a marker of human population history, were studied. In the second, the influence of S gene variants on diagnostic failure, and also how to standardise and evaluate these variants in a novel tag system were investigated. Initially (Chapter 3.1), four methods of DNA extraction have been compared to enable quick and efficient human genomic and viral DNA extraction from clotted blood. Two of these, a phenol based in-house method and Tripure isolation reagent (Roche), only achieved a low DNA yield. In contrast, QIAamp blood kit (Qiagen) and High Pure Viral Nucleic Acid kit (Roche Diagnostics) were equally efficient and sensitive. The prevalence rate of hepatitis B in the south Pacific is amongst the highest in the world. Regional immunisation programmes were introduced in four Pacific islands in 1995: Vanuatu and Fiji in Melanesia, Tonga in Polynesia and Kiribati in Micronesia (Chapter 3.2). To assess the efficacy of these programmes, sera from infants and their mothers were tested. Following immunisation there was a dramatic fall in the seroprevalence of surface antigen, especially in Fiji to 0.7% of children. However, in Vanuatu and Tonga it was 3.0% and 3.8% respectively, and occurred mostly in babies of HBeAg positive mothers denoting the importance of this mode of transmission. On the other hand, the 3.8% HBsAg prevalence in Kiribati among children of non-carrier mothers indicated that most of these infections were horizontally transmitted. This relative failure to provide satisfactory protection could reflect poor health services or insufficient neutralisation. Moreover, although the opportunity for the emergence of vaccine escape variants in these populations was high due to the presence of a considerable amount of the virus with apparently incomplete protection, there were no "a" determinant variants discovered, suggesting that these variants are relatively insignificant in this population. Some other variants were noted, but the functional significance of these remains to be determined. Geographic distribution of HBV genotypes is thought to reflect aspects of human population history. Hepatitis B virus surface antigen (HBsAg) variation from the four Pacific island locations has been analysed (Chapter 3.3). Samples were collected from unvaccinated children and adults and tested for HBsAg. At least 20 HBsAg positives from each island were amplified by PCR and sequenced. HBV isolates from C and D genotypes were identified, Genotype C predominated in Vanuatu, Fiji and Tonga while D was the dominant genotype in Kiribati. The diversity of the C genotype sample was significantly greater than that of D, consistent with a longer history of HBV infection in those islands. Strong geographic identity was evident in all populations except Tonga and Fiji, which were statistically indistinguishable. Analysis of HBV sequences from other locations will be required to fully interpret these data. HBsAg negativity does not exclude hepatitis B viraemia and HBsAg variants can be responsible for such diagnostic failures. In Chapter 3.4, we cloned 13 different HBsAg variants. Variant protein then produced in a mammalian expression system and tested using seven commercial HBsAg diagnostic assays. Of 12 variants analysed, 6 samples displayed similar reactivity to the standard HBsAg sequence in most of the assays but 6 samples, containing various mutations throughout the entire major hydrophilic region (MHR), showed reduced reactivity. Loss of cysteine at aa124 in one sample was found to influence the secretion as well as the reactivity of HBsAg in the expression system. Finally, not all assays were equally able to detect HBsAg variants implying that, to attain an acceptable level of sensitivity, the antibody repertoire of the current assays should be extended. Reduced reactivity might be due to antigenic changes or reduced particle production. To investigate the reason(s) for non-detection, supernatants derived from in-vitro expression of cloned HBsAg variants were used. We have developed an antibody capture system, using a non-HBV epitope, to standardize the amount of in-vitro expressed HBsAg protein (Chapter 3.5). Three tag systems were assessed. The successful one, influenza HA-tag, was inserted into the HBV S gene of control samples and 12 diagnostically important variants within different backbone subtypes: one aywl, four ayw2, two ayw3, and five adw2. The amount of in-vitro expressed HBsAg was then equalised in an ELISA that recognises the tag. Subsequently, the immunoreactivity of each variant was compared using three commercial HBsAg assays. We were then able to precisely attribute the diagnostic failure of the investigated variants to antigenic non-recognition and! or poor secretion of HBsAg protein. In found that single tag epitopes of up to 15 aa could be inserted at either end of HBsAg protein without affecting HBsAg reactivity. However, insertion at both ends led to a major impact on HBsAg conformation. Differentiation of second episodes of HBsAg positivity as reinfection or reactivation is a matter of debate. 5 patients who had a serological picture suggesting a second hepatitis B virus episode were studied compared to a control group of two patients who were HBsAg positive throughout with fluctuating HBeAg status. We suggest molecular criteria to distinguish between these two possibilities: number of nucleotide substitutions; number of amino acid substitutions; situation of aa changes; phylogenetic relatedness, co-incidence of mutation with immune or antiviral therapy; and genotype/ subtype shifts (Chapter 3.6). Interestingly, S gene variants were found in all 5 cases with unusual serology but in neither of the controls. Abnormal serology was therefore accompanied in all patients by rare sequences.