We have established a hepatocarcinoma cell line (LFCl2A) that produces voluminous tumors when injected into syngeneic Commentary rats. We have previously shown that when these cells were transfected with an episomal vector expressing the antisense IGFI cDNA the transduced cells partly lost their tumorigenic properties and were able to induce the regression of established hepatocarcinoma in syngeneic animals. In this paper, our aim was to determine if one could substitute the use of episomal expression vector by constructing a recombinant adenoviral vector that should be, in theory, easier to supply to humans. We have shown that, in vitro, the cells transfected as well as those infected have lost their tumorigenic properties, but in vivo the infected cells (which are no more tumorigenics) are not able to prevent tumor development.