We review the current literature on human vulnerability to environmental change and suggest that an evolutionary perspective can help integrate the insights of existing research while suggesting new hypotheses and approaches. The concept of a socially constructed adaptive landscape occupied by units at various levels of aggregation and organization -- individuals, households, organizations -- is used to organize our analysis. We argue that this framework can provide a basis for constructing dynamic stochastic models of the diverse environmental, social, economic and political factors affecting vulnerability to climate change and other natural hazards.