Mobility and access problems in rural areas of developing countries have received some attention from researchers, but this has mostly been of a qualitative nature using small-sample studies. Progressively achieving rural development objectives requires, amongst other things, a better and more quantitative understanding of the nature of the demand for mobility in rural environments, and its links with livelihoods. Rural travel demand differs from urban demand in key respects, including the existence of more restricted choice contexts. Issues of captivity and choice set formation thus need to be dealt with explicitly and carefully. The paper reports on a study undertaken in an isolated rural area within Limpopo Province in South Africa, aimed at exploring and modelling mode choice behaviour in a very constrained situation. The combined use of qualitative participatory approaches with conventional travel diary and stated choice techniques is described as a way of exploring unfamiliar aspects of rural travel behaviour. A mixed logit mode choice model is successfully calibrated, showing that rural travellers exhibit rational compensatory decision making behaviour when faced with real alternatives. However descriptive analysis suggests that, in reality, multiple alternatives are often absent or uncertain, leading to non-compensatory behaviour. This is confirmed by efforts to incorporate “soft” cutoffs to model choice set constraints for this sample.