Andrea Saltelli's paper discusses efficient methods for computing sensitivity indices in model analysis. The paper presents two new strategies for computing first and total order sensitivity indices, which are used to assess the importance of input factors in determining model output. The first strategy, based on Theorem 1, reduces computational cost by using a single set of model evaluations to compute all first and total order indices, as well as closed effect indices of order k-2. This approach reduces the number of model evaluations needed by about 50% compared to traditional methods. The second strategy, based on Theorem 2, provides double estimates of the indices, improving their accuracy. The paper also discusses the case of correlated input factors, where traditional methods are not applicable, and presents alternative approaches for sensitivity analysis. The paper includes test cases using both analytical and numerical models to demonstrate the effectiveness of the proposed methods. The results show that the new strategies significantly reduce computational cost while maintaining the accuracy of sensitivity indices. The paper concludes that these methods are particularly useful for computationally expensive models, where traditional methods would be too slow. The authors also highlight the importance of sensitivity analysis in understanding the behavior of complex models and in identifying the most influential input factors.Andrea Saltelli's paper discusses efficient methods for computing sensitivity indices in model analysis. The paper presents two new strategies for computing first and total order sensitivity indices, which are used to assess the importance of input factors in determining model output. The first strategy, based on Theorem 1, reduces computational cost by using a single set of model evaluations to compute all first and total order indices, as well as closed effect indices of order k-2. This approach reduces the number of model evaluations needed by about 50% compared to traditional methods. The second strategy, based on Theorem 2, provides double estimates of the indices, improving their accuracy. The paper also discusses the case of correlated input factors, where traditional methods are not applicable, and presents alternative approaches for sensitivity analysis. The paper includes test cases using both analytical and numerical models to demonstrate the effectiveness of the proposed methods. The results show that the new strategies significantly reduce computational cost while maintaining the accuracy of sensitivity indices. The paper concludes that these methods are particularly useful for computationally expensive models, where traditional methods would be too slow. The authors also highlight the importance of sensitivity analysis in understanding the behavior of complex models and in identifying the most influential input factors.