本研究利用實質選擇權方法及極值理論，建構不確定下最適研發投資決策模型。然後，進行數值分析與敏感度分析，探討相關參數對最適研發投資決策門檻之影響。研究結果發現：1. 考量極值分配之影響，將使研發投資決策時點延後。2. 三種不同極值分配之模糊研發投資決策門檻值以Gumbel分配模為最高，其次為Fréchet分配，最後為Weibull分配。3. 不確定程度、價值變動瞬間成長率、投資成本、模糊參數、規模參數與決策門檻值呈現正相關；而無風險利率、模糊參數、位置參數及形狀參數與決策門檻值呈現負相關。4. 本文研究模型適用實際個案資料，針對健亞生技公司之分析能說明建構理論模型之適切性。

In this study, we construct the optimal R&D investment decision-making model under uncertainty by using real options approach and extreme value theory. Then, we use the numerical analysis and sensitivity analysis to explore the influence of relevant parameters on the threshold of the optimal R&D investment decisions. The results showed that: 1. The optimal entry time of R&D investment will delay, by considering the impact of the extreme value distribution under uncertainty. 2. The fuzzy investment decision thresholds of decision model of different extreme value distribution, the Gumbel extreme value distribution is highest, the Fréchet extreme value distribution is second, the Weibull extreme value distribution is lowest. 3. The degree of uncertainty, instantaneous growth rate, investment costs, the scale parameter has the positive relation with the decision thresholds. Risk-free rate, positional parameters and shape parameters has the negative relation with the decision thresholds. 4. Our research model can apply to actual case data，the case study of the Genovate biotech company may elaborate appropriateness of the theoretical model.