Aim To quantify the vulnerability of habitats to invasion by alien plants having accounted for the effects of propagule pressure, time and sampling effort. Location New Zealand. Methods We used spatial, temporal and habitat information taken from 9297 herbarium records of 301 alien plant species to examine the vulnerability of 11 terrestrial habitats to plant invasions.A nullmodel that randomized species records across habitats was used to account for variation in sampling effort and to derive a relative measure of invasion based either on all records for a species or only its first record. The relative level of invasion was related to the average distance of each habitat from the nearest conurbation, which was used as a proxy for propagule pressure. The habitat in which a species was first recorded was compared to the habitats encountered for all records of that species to determine whether the initial habitat could predict subsequent habitat occupancy. Results Variation in sampling effort in space and time significantly masked the underlying vulnerability of habitats to plant invasions. Distance from the nearest conurbation had little effect on the relative level of invasion in each habitat, but the number of first records of each species significantly declined with increasing distance. While Urban, Streamside and Coastal habitats were over-represented as sites of initial invasion, there was no evidence of major invasion hotspots from which alien plants might subsequently spread. Rather, the data suggest that certain habitats (especially Roadsides) readily accumulate alien plants from other habitats. Main conclusions Herbarium records combined with a suitable null model provide a powerful tool for assessing the relative vulnerability of habitats to plant invasion. The first records of alien plants tend to be found near conurbations, but this pattern disappears with subsequent spread. Regardless of the habitat where a species was first recorded, ultimately most alien plants spread to Roadside and Sparse habitats. This information suggests that such habitats may be useful targets for weed surveillance and monitoring.