To design a cost-effectiveness trial, the Delta-K method newly developed provides us with a new formula for sample-size calculation. The Delta-K utilizes the difference between two cost-effectiveness ratios, qA and qB , respectively of control and experimental regimens (i.e., KD = 1/qA-1/qB ). The advantage of the method is simplicity for estimating the sample-size, naturally extending the conventional binomial formula, whereas under the assumption of "econstant" cost for each regimen. OBJECTIVES: Using multivariate analysis with the Monte-Carlo simulation for "evariable" costs, we assess the robustness of the Delta-K method in search of the probable range of sample-sizes calculated based on the formula. METHOD: A case study of cost-consequence analysis is employed to validate the design of a clinical trial, which evaluates herpes zoster treatment with Valaciclovir vs. Aciclovir. Our former investigation on this case illuminated that the sample-size for the RCT could be reduced from n=821 to n=36 in use of the Delta-K method regarding the average costs as constant. In the Monte-Carlo simulation, therefore, the three major factors of costs such as primary medication, treatment for post herpes neuralgia (PHN), and outpatient care were assumed to have a normal distribution since each factor is quantified with the mean and the standard deviation (S.D.)(e.g., for PHN; US$1504.9 + 6.1 in Valaciclovir, whereas US$2006.6 + 6.6 in Aciclovir). Under these conditions Monte-Carlo simulation was run up to 10,000 times to assess the overall distribution of sample-sizes calculated from the formula. RESULTS: The runs of 10,000 times resulted in the distribution with the sample-size of 35.2 + 0.8(the mean + 2S.D.) and the range of (33.7, 37.0) = (min, max), which is far smaller than the original sample-size n = 821. CONCLUSION: We confirmed sufficient robustness of the Delta-K method in a case study even if medical costs vary against the assumption of constant cost. It suggests the usefulness of the new method.