When dealing with a computer model which simulates real phenomena, global sensitivity analysis techniques aim to apportion the model's output uncertainty to uncertainty in its inputs. The results of this can then be used for model calibration, model validation, and decision-making processes, i.e., any processes where it is useful to know which variables contribute most to output variability. In many works, much attention has been paid to screening techniques [111] and variance-based sensitivity measures, also known as Sobol' indices [442]. Indeed, numerical model builders and users have shown great interest in tools like these, which take full advantage of the advent of high-powered computing and numerical methods; see [187, 110, 126] for industrial and environmental applications. Several texts [414, 415, 111, 127, 110, 395] have covered these topics rather extensively. However, none of these describes the many recent technical advances in sensitivity analysis (SA). This is the main goal of the book: to provide a unified view of most of the new theoretical and algorithmic results in and around SA while keeping an eye on applications of the methods described. For a more extensive view of practical issues related to SA, the reader could refer to the recent position paper of Razavi et al. [395].