OBJECTIVE Missense mutations in HIV-1 reverse transcriptase are frequently selected in response to therapy; we examined whether silent mutations were also selected for by HIV therapy. METHODS Retrospective, observational analysis. Biochemical assays. METHODS A comparison of the reverse transcriptase gene, from antiretroviral- naive (N = 812) and experienced individuals (N = 2212), reveals two silent mutations (K65K and K66K) that are strongly associated with treatment experience. To assess reverse transcription efficiency, steady-state kinetic assays were carried out using recombinant purified HIV-1 reverse transcriptase and a series of synthetic RNA/DNA template/primer substrates. The RNA templates spanned codons 60-77 in the reverse transcriptase and included different combinations of mutations at codons 65, 66, 67, and 70. RESULTS Silent AAG mutations (or mixtures) at reverse transcriptase codons 65 and/or 66 were observed in 812 samples from 351 patients and 2129 samples from 829 patients, respectively. In clade B samples, there was a very strong relationship between the silent mutations and the thymidine analogue mutations, in particular D67N. Steady-state kinetic experiments demonstrated that HIV-1 reverse transcriptase exhibited a strong tendency to pause and/or dissociate at codons 65 and 66 on RNA templates that contained the D67N and K70R mutations. However, when the K66 or K66 AAA to AAG mutations were added to the background of the 67 and 70 mutational changes, these pausing and/or dissociation events were largely alleviated. CONCLUSIONS Silent mutations at codons 65 and/or 66 are strongly coselected with thymidine analogue mutations. These data provide the first evidence for an RNA-level mechanism of direct relevance to drug resistance.