In the last meeting of ISPOR, we presented a paper that showed the actual cost-effectiveness of treatments in practice can be reduced by the influence of statistical errors such as type II (ß) and type III (?). OBJECTIVE: Based on that research, this paper developed a further theoretical framework for such estimation which provides a new method for power analysis of a clinical trial and proper translation of its outcomes in terms of cost-effectiveness. METHODS: Consider a statistical model of a clinical trial for two regimens: a new treatment TA and TB. Under the traditional two-sample binomial analysis, we theoretically expanded it into power and sample sized calculations with the cost-effectiveness ratios ?A and ?B, respectively, for the regimens TA and TB. Furthermore, we assume that the truth is TA is more cost-effective than TB. Then, the binomial analyses were generalized under the decision-analytic model that can reflect a reality in practice with three alternatives according to no error, ß and ? errors possibly contained in the statistical conclusion of the clinical trials for TA and TB. RESULTS: Two formulae were developed mathematically to incorporate cost-effectiveness; one is an expansion of the traditional formula for sample-size calculation with the cost-effectiveness ratios ?A and ?B, and the other a more realistic one with ?A and ?B, and the modified cost-effectiveness ratio adjusted by the statistical ß and ? errors using ?A and ?B. CONCLUSIONS: The newly developed formulae can offer a more theoretically sound description to design a clinical trial beyond the traditional techniques.