The hepatitis B virus (HBV) is classified into distinct genotypes A-H that are characterized by different progression of hepatitis B and sensitivity to interferon treatment. Previous computational genotyping methods are not robust enough regarding HBV dual infections with different genotypes. The correct classification of HBV sequences into the present genotypes is impaired due to multiple ambiguous sequence positions. We present a computational model that is able to identify and genotype inter- and intragenotype dual infections using population-based sequencing data. Model verification on synthetic data showed 100 % accuracy for intergenotype dual infections and 36.4 % sensitivity in intragenotype dual infections. Screening patient sera (n = 241) revealed eight putative cases of intergenotype dual infection (one A-D, six A-G and one D-G) and four putative cases of intragenotype dual infection (one A-A, two D-D and one E-E). Clonal experiments from the original patient material confirmed three out of three of our predictions. The method has been integrated into geno2pheno([hbv]), an established web-service in clinical use for analysing HBV sequence data. It offers exact and detailed identification of HBV genotypes in patients with dual infections that helps to optimize antiviral therapy regimens. geno2pheno([hbv]) is available under http://www.genafor.org/g2p_hbv/index.php.