Rather than attempt to write a balanced or complete overview of the application of GIS to archaeology (which would inevitably end up being didactic and uncritical) this article sets out to present a discursive and contentious position with the deliberate aim of stimulating further debate about the future role of GIS within the discipline. To this end, existing applications of GIS to archaeology are reviewed, concentrating on two areas of application, predictive modelling and visibility analyses, and on their wider disciplinary context. It is argued that GIS cannot be simplistically held to have been a 'good thing' or a 'bad thing' for archaeology, but rather that these different application areas may be analysed separately and found to have quite different qualities. Although they are in no sense alternatives to one another, the areas of predictive modelling and visibility analysis can be seen to represent quite different agendas for the development of an archaeology of space and/or place. The development of correlative predictive models is considered first, both from the perspective of explanation and of cultural resource management. The arguments against predictive modelling as a means of explanation are rehearsed and it is found to be over-generalising, deterministic and de-humanised. As a consequence, it is argued that predictive modelling is now essentially detached from contemporary theoretical archaeological concerns. Moreover, it is argued to be an area with significant unresolved methodological problems and, far more seriously, that it presents very real dangers for the future representativity of archaeological records. Second, the development of GIS-based visibility analysis is reviewed. This is also found to be methodologically problematic and incomplete. However, it is argued that visibility studies — in direct contrast to predictive modelling — have remained firmly situated within contemporary theoretical debates, notably about how human actors experience places (phenomenology) and perceive their surroundings (cognition). As such, it is argued that visibility analysis has the potential to continue to contribute positively to the wider development of archaeological thinking, notably through laying the foundations of a human-centred archaeology of space. The paper concludes by qualifying the claim that there is a 'hidden agenda' for archaeological applications of GIS, particularly by making it clear that this does not imply an attempt to distort the discipline. Instead, this is explained in terms of institutional and disciplinary inertia that should be addressed through greater debate and communication over these issues.