Summary. There is little recent data of the seroprevalence of hepatitis B in Australia. We have surveyed a large cohort of endoscopy patients attending a teaching hospital in central Sydney, and related the presence of hepatitis B virus (HBV) markers with putative risk factors for exposure using the SAS statistical package. Of the 2115 patients tested: 2.1% (45/2115) were HBV surface antigen positive, 0.75% (14/2115) viraemic, 9.5% (200/2115) anti-HBs and anti-HBc positive, 20.1% (430/2115) vaccinated (anti-HBs only) and the remaining 70% were susceptible. The adjusted OR of HBV infection was significantly increased in patients who had been diagnosed with human immunodeficiency virus (36.3-fold), born in Asia or Pacific islands (12.4-fold), born in North Africa, Middle East & Mediterranean countries (6-fold) or born abroad elsewhere in the world (2.7-fold), had household contact with someone diagnosed with hepatitis between 1980 and 1990 (3.9-fold), injected drugs between 1980 and 1990 (4.4-fold), resided in a military establishment for 3 months (2.3-fold) or in a hospital for 3 months (2.2-fold), never been vaccinated for hepatitis B (2.8-fold), received blood transfusion due to an accident and/or a haemorrhage (1.92-fold) and finally been a male gender (1.59-fold). The prevalence of HBV in this hospital population was higher than predicted on the basis of notifications to the passive surveillance scheme. Most HBV patients had multiple risk factors for infection, but the hierarchy of odds ratios provides a rational basis for targeted programmes to identify asymptomatic HBV carriers who might benefit from treatment.