Understanding and predicting invasion of exotic species presents one of the greatest challenges in ecology today. In fragmented landscapes, habitat connectivity may play a key role in determining the locations that are most susceptible to invasion. This research examines exotic plant invasion in forest patches in two national battlefields in Maryland.  Our goal was to identify the relationship between habitat patch connectivity and invasion of exotic species. We used graph theory to quantify habitat connectivity in a variety of ways, using local metrics such as degree (the number of patches directly connected to a patch) as well as landscape metrics such as betweenness and closeness, which measure spatial pattern at a larger scale. Connectivity was measured for a variety of dispersal distances, and distance between patches was measured using both Euclidean distance and cost-path distance. Plants were grouped according to species traits such as life form, seed size, and dispersal mechanism. Results suggest that configuration of habitat patches can either facilitate or inhibit spread of invasive species, depending on their dispersal abilities. We found that plant communities in isolated habitat patches differed from plant communities in more connected patches. The best connectivity metric for predicting invasion varied according to species traits, but in general the landscape-scale metrics predicted invasion better than local-scale metrics. This graph theory approach holds great potential for predicting future invasions as well as for focusing management and eradication efforts.