Agent-based modelling and simulation of survival and settlement behaviour via interactive visualisation could potentially become a useful technique for generating new knowledge in those areas that sparse information, acquired through traditional methods, does not allow researchers to make informed decisions about past behaviour. This is particularly important following the development of remote sensing technologies that, in marine environments, are permitting novel exploration of previously inaccessible historic landscapes. This article explores and develops an agent-based model for basic survival and settlement behaviour for Mesolithic communities based within a marine palaeolandscape. It discusses the issues regarding how agents can be created to react to resource and environmental needs and limitations. The methodological study examines individual agent behaviour and sets the foundation for future, more complex scenarios that span large spatio-temporal landscapes including the North Sea and European coastal shelves. The article also considers the key technical challenges that must be met if large complex scenarios emerge in which the modelling of interaction between vegetation, animals and human groups becomes a priority.