BACKGROUND: Consensus on the interpretation of mutations in the human immunodeficiency virus (HIV)-1 reverse transcriptase (RT) gene that predict the response to didanosine treatment is needed. METHODS: Baseline HIV-1 RT genotypes and 12-week virological outcomes for patients undergoing didanosine-containing salvage regimens were extracted from prospective studies. Existing didanosine genotypic-resistance interpretation rules were validated in the entire-patient data set. Mutations were given weighted positive or negative scores according to their coefficient of correlation with virological response in a derivation set. The score resulting from the algebraic sum of the mutations was then validated in an independent data set. RESULTS: A total of 485 patients were analyzed. The didanosine-resistance scores derived from the Jaguar and Gesca studies predicted virological outcome. The best correlation with response was found with the derived score (M41L x 2) + E44D/A/G + T69D/S/N/A + (L210W x 2) + T215Y or revertants + L228H/R - D123E/N/G/S, by use of which viruses were categorized as being susceptible (score < or =0), as having intermediate resistance (1-3), and as being resistant (> or =4) to didanosine. In the validation set, the adjusted mean difference in 12-week virological response was +0.34 log(10) copies/mL (95% confidence interval, +0.11 to +0.57; P=.004) per higher resistance category. Correlation with virological response constantly outperformed that obtained with the previous interpretation. CONCLUSION: The improved genotypic-resistance interpretation score can be applied to better guide the use of didanosine in treatment-experienced individuals.