OBJECTIVES: Cost-effectiveness estimates for new drugs are generally derived from limited data on product efficacy and likely treatment patterns. We evaluated the base-case cost-effectiveness, from a US perspective, of ritonavir-boosted darunavir (TMC114/r; darunavir/r), a novel protease inhibitor, plus an optimized background regimen (OBR) compared with currently available protease inhibitors plus OBR, for treating HIV infection in treatment-experienced adults. We then subjected these estimates to sensitivity and variability analyses, to assess the effects of uncertainty in the base-case estimates. METHODS: A Markov model with 3-month cycles was developed to follow patients through six health states defined by CD4+ cell-count ranges. Transition probabilities were calculated from clinical trial data. Cost (2005 US dollars), utility, and mortality data were estimated from published US sources. The 5-year incremental cost-effectiveness ratio (ICER) was calculated for the base case. The base case was tested with one-way and probabilistic sensitivity analyses (PSA) as well as for alternative practice patterns, population and model characteristics (variability analyses). RESULTS: The base-case 5-year ICER for darunavir/r was $4,657/QALY. The one-way sensitivity analysis results ranged from -$12,086/QALY (darunavir/r is dominant) to +$24,899/QALY. Results were most sensitive to uncertainty about durations of CD4+ increase and stability, rates of CD4+ change, and non-antiretroviral healthcare costs. By PSA, the probability of the ICER being below the commonly accepted threshold of $50,000/QALY was 95.5%. The variability analysis resulted in ICERs between -$42,029 (darunavir/r is dominant) and +$33,625. The lower bound occurred when tipranavir was the assumed comparator, while the upper bound occurred when the model time horizon was extended to lifetime. CONCLUSION: The impact of parameter uncertainty assessed by sensitivity analysis on the estimated ICER was less than the impact of variability. Each type of analysis contributes to a fuller understanding of the uncertainty in base-case cost-effectiveness estimates. After analyses of uncertainty, darunavir/r remains cost-effective.