Hepadnaviruses, including hepatitis B virus (HBV), a highly relevant human pathogen, are small enveloped DNA viruses that replicate via reverse transcription. All hepadnaviruses display a narrow tissue and host tropism. For HBV, this restricts efficient experimental in vivo infection to chimpanzees. While the cellular factors mediating infection are largely unknown, the large viral envelope protein (L) plays a pivotal role for infectivity. Furthermore, certain segments of the PreS domain of L from duck HBV (DHBV) enhanced infectivity for cultured duck hepatocytes of pseudotyped heron HBV (HHBV), a virus unable to infect ducks in vivo. This implied a crucial role for the PreS sequence from amino acid 22 to 90 in the duck tropism of DHBV. Reasoning that reciprocal replacements would reduce infectivity for ducks, we generated spreading-competent chimeric DHBVs with L proteins in which segments 22-90 (Du-He4) or its subsegments 22-37 and 37-90 (Du-He2, Du-He3) are derived from HHBV. Infectivity for duck hepatocytes of Du-He4 and Du-He3, though not Du-He2, was indeed clearly reduced compared to wild-type DHBV. Surprisingly, however, in ducks even Du-He4 caused high-titered, persistent, horizontally and vertically transmissable infections, with kinetics of viral spread similar to those of DHBV when inoculated at doses of 10(8) viral genome equivalents (vge) per animal. Low-dose infections down to 300 vge per duck did not reveal a significant reduction in specific infectivity of the chimera. Hence, sequence alterations in PreS that limited infectivity in vitro did not do so in vivo. These data reveal a much more complex correlation between PreS sequence and host specificity than might have been anticipated; more generally, they question the value of cultured hepatocytes for reliably predicting in vivo infectivity of avian and, by inference, mammalian hepadnaviruses, with potential implications for the risk assessment of vaccine and drug resistant HBV variants.