The use of antisense oligodeoxyribonucleotides (ODNs) to inhibit translation of mRNAs promises to be an important means of controlling gene expression and disease processes. ODNs are about 20 nucleotides long, so hundreds of possible targets are available in a given mRNA. An elusive goal has been to efficiently predict the best in vivo antisense target without having to study a large pool of possible ODN sequences for each mRNA. It would be a breakthrough if ODN selection could be accurately guided by the application of sequence specific parameters to an mRNA sequence. The selection of the best ODN sequence is complicated since cellular uptake, conditions at the mRNA target site, non-sequence-specific effects, sequence redundancy, and mRNA secondary structures are difficult to predict. Thermodynamic parameters for nearest-neighbor (dimer) duplex stabilities, from in vitro studies, have not been adequate predictors of in vivo hybridization. The methodology of this application shows that it is possible to obtain parameters for in vivo motifs, which are defined as combinations of next-nearest-neighbors, that are correlated with efficient antisense targeting. These parameters can be used to identify mRNA sequences that are binding sites for effective antisense ODNs. Next-nearest-neighbor nucleotide parameters can be derived directly from cell culture inhibition data so that in vivo conditions are taken into account.