We are developing triple helix-forming oligonucleotides (TFOs) as gene targeting reagents in living mammalian cells. We have described psoralen-linked TFOs with 2'-O-methyl and 2'-O-(2-aminoethyl) (2'-AE) substitutions that are active in a gene knockout assay in cultured cells. The assay is based on mutagenesis by psoralen, a photoactive DNA cross-linker. Previous work showed that TFOs with three or four 2'-AE residues were disproportionately more active than those with one or two substitutions. Here we demonstrate that for optimal bioactivity the 2'-AE residues must be clustered rather than dispersed. We have further characterized bioactive and inactive TFOs in an effort to identify biochemical and biophysical correlates of biological activity. While thermal stability is a standard monitor of TFO biophysical activity, we find that T(m) values do not distinguish bioactive and inactive TFOs. In contrast, measurements of TFO association rates appear to correlate well with bioactivity, in that triplex formation occurs disproportionately faster with the TFOs containing three or four 2'-AE residues. We asked if extending the incubation time prior to photoactivation would enhance the bioactivity of a TFO with a slow on rate relative to the TFO with a faster association rate. However, there was no change in bioactivity differential. These results are compatible with a model in which TFO binding in vivo is followed by relatively rapid elution by cellular functions, similar to that described for transcription factors. Under these circumstances, TFOs with faster on rates would be favored because they would be more likely to be in triplexes at the time of photoactivation.