Background: The human immunodeficiency virus type I (HIV-I) regulatory protein, Nef, is an attractive vaccine target because it is involved in viral pathogenesis, is expressed early in the viral life cycle and harbors many T and B cell epitopes. Several clinical trials include gene-based vaccines encoding this protein. However, Nef has been shown to transform certain cell types in vitro. Based on these findings we performed a long-term toxicity and immunogenicity study of Nef, encoded either by Modified Vaccinia virus Ankara or by plasmid DNA. BALB/c mice were primed twice with either DNA or MVA encoding Nef and received a homologous or heterologous boost ten months later. In the meantime, the Nef-specific immune responses were monitored and at the time of sacrifice an extensive toxicological evaluation was performed, where presence of tumors and other pathological changes were assessed. Results: The toxicological evaluation showed that immunization with MVAnef is safe and does not cause cellular transformation or other toxicity in somatic organs. Both DNAnef and MVAnef immunized animals developed potent Nef-specific cellular responses that declined to undetectable levels over time, and could readily be boosted after almost one year. This is of particular interest since it shows that plasmid DNA vaccine can also be used as a potent late booster of primed immune responses. We observed qualitative differences between the T cell responses induced by the two different vectors: DNA-encoded nef induced long-lasting CD8+ T cell memory responses, whereas MVA-encoded nef induced CD4+ T cell memory responses. In terms of the humoral immune responses, we show that two injections of MVAnef induce significant anti-Nef titers, while repeated injections of DNAnef do not. A single boost with MVAnef could enhance the antibody response following DNAnef prime to the same level as that observed in animals immunized repeatedly with MVAnef. We also demonstrate the possibility to boost HIV-I Nef-specific immune responses using the MVAnef construct despite the presence of potent anti-vector immunity. Conclusion: This study shows that the nef gene vectored by MVA does not induce malignancies or other adverse effects in mice. Further, we show that when the nef gene is delivered by plasmid or by a viral vector, it elicits potent and long-lasting immune responses and that these responses can be directed towards a CD4+ or a CD8 + T cell response depending on the choice of vector.