High transgene stabilities of 1 year and more have been reported in immunodeficient hosts after adenovirus-mediated gene transfer. Transgene persistence of this duration could be due to inherently high stability of the episomal viral vector DNA. An alternative explanation would be limited 'autoreplication' of transgenic vector DNA, just sufficient to counteract slow but continuous degradation within the host cells. Autoreplication could occur in the absence of any production of infectious virus particles, based on residual activity of the adenoviral DNA replication system only. To test this hypothesis, a series of DNA metabolic labeling studies in non-permissive cells cultures transfected with different vectors was conducted. Due to extensive E1 region deletions none of the vectors was able to produce viral progeny in non-permissive cells. Vectors fell into two categories, however, with respect to their autoreplication potential. Neosynthesis of vector DNA in non-permissive vector-transfected cells was readily detectable in 'type A', but not in 'type B' vectors. In addition to their different transgene expression cassettes, vector DNA sequencing showed a less extensive E1 deletion in type A (nucleotides 453-3333 of wild-type virus) as compared to type B vectors (nucleotides 325-3523). Autoreplication was also associated with high transcriptional activity of several viral genes (E1B-14k, adenoviral DNA polymerase, single-strand DNA-binding protein, E4-25k), in contrast to type B vectors. In addition to these 'wild-type' transcripts, 'irregular' recombinant transcripts were detected in autoreplication vectors which contained the transgenic cDNA in conjunction with adenoviral vector sequences. Exogenous or cryptic promotors may (under certain conditions) enhance the transcriptional activity of a vector in such a way that autoreplication occurs. Conditions determining the level of transcriptional enhancement (extent of E1 deletion, type of promoter and transgene, etc) need to be further defined before rational design of adenovectors with high autoreplication capacity becomes possible. In summary, we have shown autoreplication to be a novel feature of certain E1-deleted adenovectors with likely relevance for their stability in vivo, but also with possibly adverse consequences for target cell function or vector immunogenicity. Full characterization of adenoviral vector systems should therefore include a description of their autoreplication capacity.