Efficient gene transfer with adenoviral type 5 (Ad5) vectors depends on the initial attachment of their fiber, which binds the coxsackie-adenovirus receptor (CAR), and their subsequent internalization, mediated by the interaction of viral penton base with target cell alphav integrins. We previously demonstrated that human leukemic cells lack these receptors and are therefore resistant to Ad5 transduction, limiting efforts to genetically modify these cells. Human leukemic blasts are, however, susceptible to transduction with an adenovector made CAR independent by substitution of a chimeric Ad5/35 fiber [Yotnda et al. (2001). Gene Ther. 8, 930-937]. Other receptors can also be targeted with recombinant ligand moieties incorporated into adenovirus fiber. We have determined which of these fiber-modified adenovectors is most effective at modifying human primary leukemia cells, and lines. We used a replication-incompetent Ad5-beta-gal vector, in which the Ad5 fiber was replaced with fiber from various adenovirus serotypes (Ad35 and Ad11), or modified either with variable length polylysine (K4, K7, K21) or RGD-4C peptide. All the modified fiber vectors transduced primary leukemia cells and cell lines more efficiently than Ad5. Polylysine-substituted Ad5F/K21 and peptide-modified Ad5F/RGD vectors were most effective overall (up to 100% efficiency), whereas Ad5F/RGD was the most effective at transducing B cell acute lymphoblastic leukemia cells (90% efficiency). Ad5F/K21 and Ad5F/RGD should be of value for the genetic modification of human primary leukemia cells.