Chronic infection with hepatitis B virus (HBV), a small DNA virus that replicates by reverse transcription of a pregenomic (pg) RNA precursor, greatly increases the risk for terminal liver disease. RNA interference (RNAi) based therapy approaches have shown potential to overcome the limited efficacy of current treatments. However, synthetic siRNAs as well as small hairpin (sh) RNAs expressed from non-integrating vectors require repeated applications; integrating vectors suffer from safety concerns. We pursue a new concept by which HBV itself is engineered into a conditionally replicating, wild-type HBV dependent anti-HBV shRNA vector. Beyond sharing HBV's hepatocyte tropism, such a vector would be self-renewing, but only as long as wild-type HBV is present. Here, we realized several important aspects of this concept. We identified two distinct regions in the 3.2 kb HBV genome which tolerate replacement by shRNA expression cassettes without compromising reverse transcription when complemented in vitro by HBV helper constructs or by wild-type HBV; a representative HBV shRNA vector was infectious in cell culture. The vector-encoded shRNAs were active, including on HBV as target. A dual anti-HBV shRNA vector delivered into HBV transgenic mice, which are not susceptible to HBV infection, by a chimeric adenovirus-HBV shuttle reduced serum hepatitis B surface antigen (HBsAg) up to ˜4-fold, and virus particles up to ˜20-fold. Importantly, a fraction of the circulating particles contained vector-derived DNA, indicating successful complementation in vivo. These data encourage further investigations to prove antiviral efficacy and the predicted self-limiting vector spread in a small animal HBV infection model.