The usual course of drug discovery begins with the demonstration of compound activity in cells and, usually, a lower level of activity in animals. Successive rounds of drug design may result in a compound with sufficient activity in animals to justify clinical trials. The basic endpoints of therapeutic oligonucleotide experiments include target antigen reduction, target messenger reduction and inhibition of transformed cell proliferation or viral replication. However, one should expect oligonucleotides to exhibit pleiotropic behaviour, as do all other drugs. In an animal oligonucleotides will necessarily bind to and dissociate from all macromolecules encountered in the blood, in tissues, on cell surfaces and within cellular compartments. Contrary to expectations, oligonucleotides designed to be complementary to certain transcripts have sometimes been found moderately effective in cell-free extracts, more effective in cell culture and most effective in animal models. If greater potency against standard endpoints is reported in mouse models than was observed in cell culture, critical examination must consider alternate modes of action in animals that may not apply in cell culture. This counterintuitive paradox will be examined, based on studies of Ha-ras expression in bladder cancer, Ki-ras expression in pancreatic cancer, erbB2 expression in ovarian cancer and c-myc expression in B cell lymphoma.