Detection of hepatitis B virus (HBV) DNA in serum allows monitoring of HBV replication and assessing responses to antiviral treatment. HBV DNA quantification measures virus replication and can be used as a prognosis indicator of liver disease and an index of response to antiviral drugs. The aim of this study was to compare the performances of three HBV DNA detection and/or quantification techniques for assessing HBV replication. Three hundred unselected sera with a request for HBV DNA detection and quantification were tested with a molecular hybridisation technique without amplification (Digene Hybrid-Capture, Murex Diagnostics Ltd), a signal amplification assay based on branched DNA technology (Quantiplex HBV DNA, Chiron diagnostics), and an 'in-house' qualitative, non quantitative target amplification assay based on the polymerase chain reaction (PCR) with primers located in the S gene of the HBV genome. Hybrid-capture and branched DNA gave concordant results in 278 cases (93%). In the 128 samples positive by both assays, DNA titres in pg/ml were related significantly (r = 0.70, P < 0.0001). but branched DNA titres increased more rapidly than hybrid-capture titres when the amount of HBV DNA in the sample increased. Twenty-two sera (7%) were negative by hybrid-capture, but positive in branched DNA (detection rate gain: 15%). In these 22 patients, DNA titres were low, HBsAg was present in all instances and alanine aminotransferase activity was elevated in 18 patients (82%); HBeAg was present in seven patients (32%) and anti-HBe antibodies in 18 patients (82%); liver biopsy, undertaken in 18 patients, revealed chronic active hepatitis in all instances, associated with cirrhosis in eight cases. Qualitative, non-quantitative HBV DNA PCR was positive in 75 (50%) of the 150 hybrid-capture-negative, branched DNA-negative samples, including a significant proportion of patients without evidence of ongoing HBV-related liver disease. The results show that in general, the branched DNA assay detects HBV DNA in more patients than hybrid-capture and that this improved detection rate is relevant clinically and genome equivalents/ml are preferred to pg/ml to quantify HBV DNA in clinical specimens and finally qualitative, non-quantitative polymerase chain reaction can detect HBV DNA in patients without evidence of active HBV-related liver disease. This study emphasizes the need for more sensitive, university standardised quantitative HBV DNA assays and for the definition of clinically relevant cutoffs with these assays.