Mathematical models are developed to approximate engineering, physical, environmental, social, and economic phenomena of various complexities. To ensure a correct use of a model, an appreciation of the relationship between (model) input factors and (model) output is of fundamental importance. The process aiming to understand how a model behaves in response to changes in its inputs is known as Sensitivity Analysis. Sensitivity Analysis applies to a variety of settings and fields. A large number of different methodologies, developed by scientists from a wide spectrum of disciplines, can be found in the literature. In September 1995, the Joint Research Centre (Ispra) organised the first international symposium on Sensitivity Analysis of Model Output (SAM095). SAM095 was successful in bringing together researchers from different scientific backgrounds, involved in theoretical and practical aspects of sensitivity analysis, and offered a review of the current state of this art. The second international symposium, SAM098, is organised by the Sensitivity Analysis group at the Institute for Systems Informatics and Safety, JRC and by the University of Venice. Over 150 scientists from different disciplines and countries contribute to SAM098 by presenting achievements, either in theoretical developments or practical applications. The final programme of the symposium includes nearly 50 oral presentations, organised into 13 thematic sessions, and 2 poster sessions, with over 30 contributions. An extended abstract of each presentation and poster is contained in the present volume. The number of participants and the variety of works presented at SAM098 confirm the growing importance of Sensitivity Analysis in all settings and fields where modelling applies. Readers of this volume might be interested in a special issue'" based on SAM098, where some selected topics addressed at the symposium are dealt with more in depth. The topics include SA applications to expert elicitation procedures within the Bayesian context, efficient input-output model representations, variance decomposition methods and their applications, hybrid approaches to SA, identification of model behaviour in the vicinity of extremes, generalised sensitivity analysis (related to Monte Carlo filtering), advanced model representations such DBM (Data Based Mechanistic) modelling, local automated SA (to compute large sensitivity matrices) and uncertainty in performance assessment.
