Assessing vulnerability to climate change can help policymakers in incorporating climate futures in planning and in better allocating adaptation resources. Indicator Based Vulnerability Assessment (IBVA) has been widely used because it is relatively simple to design, implement and communicate. However, this approach faces significant difficulties from conceptual, theoretical and methodological points of view. A number of assumptions are typically made in methods used for aggregation of indicators—a linear, monotonic relationship between indicator and vulnerability; complete compensation between indicators; precise knowledge of vulnerable systems by stakeholders who provide input data for the assessment exercise—none of which usually hold in reality. Following a meta-analysis of the IBVA literature, the thesis proposes a) a general mathematical framework for vulnerability assessment that better identifies sources of uncertainty and non-linearity; b) a new IBVA assessment methodology, and associated computer tool, based on a pair-wise outranking approach borrowed from decision science; the methodology can represent various sources of uncertainty, different degree of compensation between indicators and different types of non-linearity in the relationship between indicators and vulnerability and; c) a system dynamics model, integrated in the above framework, for studying vulnerability of infrastructure systems and better representing the mechanistic interdependency of their components. These methods are applied to a real-life assessment of the vulnerability to sea-level rise of communities and infrastructure systems in Shoalhaven, south of Sydney, at local scale. The assessment is conducted in collaboration with the Shoalhaven council and includes an analysis of the sensitivity of vulnerability rankings to community preferences. In addition, the effect of using an outranking framework on the way vulnerability is conceptualized by stakeholders is critically appraised.