Effective management of introduced species requires the early identification of species that pose a significant threat of becoming invasive. To better understand the invasive ecology of species in New England, USA, we compiled a character data set with which to compare non-native species that are known invaders to non-native species that are not currently known to be invasive. In contrast to previous biological trait-based models, we employed a Bayesian hierarchical analysis to identify sets of plant traits associated with invasiveness for each of three growth forms (vines, shrubs, and trees). The resulting models identify a suite of ‘invasive traits’ highlighting the ecology associated with invasiveness for each of three growth forms. The most effective predictors of invasiveness that emerged from our model were ‘invasive elsewhere’, ‘fast growth rate’, ‘native latitudinal range’, and ‘growth form’. The contrast among growth forms was pronounced. For example, ‘wind dispersal’ was positively correlated with invasiveness in trees, but negatively correlated in shrubs and vines. The predictive model was able to correctly classify invasive plants 67% of the time (22/33), and non-invasive plants 95% of the time (204/215). A number of potential future invasive species in New England that deserve management consideration were identified.