Across the globe, invasive alien plant species are a major threat to biodiversity and agricultural industries.Invasive plant spread is an ongoing process that plays out at different spatiotemporal scales.When an invasive 'weed' emerges in a new country, region or landscape, there is often a distinct lag phase during which population increase and invasive spread is slow.Rapid response by biosecurity managers to new incursions may bring the greatest opportunities with regard to facilitating eradication or effective containment as well as minimizing long-term control costs and negative impacts.An important consideration for allocating finite management resources is the spatial distribution of invasion risk.However, hard data is rarely available to assess risk, and decisions are consequently made under considerable uncertainty.We developed a flexible, spatially-explicit methodology for rapidly modelling invasion risk using Bayesian networks and Geographic Information Systems (GIS).In this well-established modelling framework, the key processes of invasion and the major risk variables affecting these processes are represented as a Bayesian network.GIS analysis is used to link each risk variable contained in the model to one or several spatial data proxies and to map model predictions.Here, we present a simplified implementation that can be used rapidly and under limited knowledge, using the case study of Mexican bean tree (Cecropia spp.) in North Queensland.We modelled two aspects of invasion risk: (a) the potential, time-independent environmental suitability of a site for plant establishment and persistence, and (b) the actual susceptibility of suitable sites to propagule introduction from source infestations recorded during annual 'detection periods'.We derived risk variables and spatial data proxies via desktop research and used simple mathematical functions to model the interplay between risk variables, invasion processes (establishment, persistence and propagule pressure), and invasion risk (suitability and susceptibility).We also evaluated the ability of our model to accurately predict the observed spatial progression of the Cecropia spp.invasion in North Queensland by validating mapped annual susceptibility against subsequently observed infestations.The majority of North Queensland's Wet Tropics was predicted to be potentially highly or moderately suitable for Cecropia spp.invasion.Actual susceptibility to invasion in a given fruiting season is more restricted due to dispersal constraints from source infestations.In each evaluated detection period, modelled susceptibility was a good predictor of actual Cecropia spp.detections in subsequent years.Annual susceptibility maps may be used for allocating surveillance and containment resources around existing infestations.Low-frequency long-distance dispersal or human-mediated translocations were not captured in the model.Further automation of risk factor / spatial proxy selection and data pre-processing may make our methodology for rapidly modelling invasion risk accessible to biosecurity agencies that typically have significant field knowledge and GIS capacities but may have limited expertise in applying complex modelling methods.