Weed risk assessment has become an accepted methodology for examining the likelihood and consequence of a plant species becoming invasive outside of its native range. Weed risk assessment draws upon biological and ecological information to estimate the likelihood and magnitude of the threats posed by introducing non-indigenous plants. In geographical terms, this has traditionally been understood as within a new country following importation of plant material. However, recent risk assessment development has focused more specifically on intracountry risk posed by already-present invasive plants and is referred to as post-border weed risk management. This form of assessment calls for fine-scale predictions of invasive species habitat suitability. This study applies some of the more popular and widely available habitat prediction models that represent a variety of different statistical approaches (linear regression, logistic regression, Bayesian probability, Classification and Regression Trees, Genetic Algorithm for Rule-set Production) to a single invasive plant, the vertebrate-dispersed, fleshy fruited European olive (Olea europaea L.) in southern Australia. The relationships between the dependant (O. europaea distribution) and independent (soil and climate) variables are used in the models to produce predictive maps for each model. Accuracy was calculated for each model output as well as a combined surface to examine whether recent calls for ensemble modelling of distributions produces improved predictions. Overall, the combined prediction demonstrated superior accuracy compared to any individual model outputs. The combined outputs can be likened to mapped gradations of predicted habitat suitability. The type of output produced in this study should form a critical component of post-border weed risk management but more importantly, the methodology will add to this important discipline.