Assisted colonisation, one form of species translocation, has been proposed as a tool for helping species to track suitable conditions in a changing climate. There are considerable practical challenges associated with it, including predicting where to place translocated individuals. This problem may be particularly big for small and immobile species, where small-scale micro-environmental conditions de-couple them from environmental conditions as projected in large-scale climate models. To investigate this problem we developed a survey-based model to predict the occurrence of our target species, the fruticose terricolous arctic-alpine lichen, Flavocetraria nivalis, within the Cairngorm Mountains. We then undertook an experimental translocation of this species. A second model, using variables that were significant in the survey-based model, was only fair at predicting the initial pattern of survival at the recipient site. However, model fit of the translocation survival model improved over time as the distribution of surviving individuals more accurately reflected the distribution of suitable environmental conditions. In addition, model predictive power increased with the addition of data on micro-climatic conditions at recipient plots. Synthesis and applications. Our results demonstrate that, for species which respond strongly to local environmental conditions, are immobile and, to some extent, decoupled from larger-scale climates, it may be difficult to build a priori accurate predictive models of habitat suitability. In these cases, a combination of modelling and expert judgement, along with the movement of substantial numbers of transplants, may be the appropriate options for maximising the success of assisted colonisation.