Understanding establishment and spread of non-native plants is important in the face of a homogenizing global flora. While many studies focus on successful, invasive species, fewer have studied failed plant introductions. Until the early 1900s, large quantities of ship ballast, often containing foreign plant propagules, were deposited in New Jersey (USA). The resulting ballast flora is documented in extensive herbarium records, providing us a unique opportunity to analyze successes and failures of novel plant species introductions. We used digitized specimens from 75 herbaria to study 264 non-native species introduced into New Jersey through 19th century ballast deposition. We used spatial (density-based clustering; HDBSCAN) and temporal analyses of species retention and geographic spread to quantify disappearance rate, survival, and dispersion through time and define trajectory groups. Four distinct trajectory groups were identified: waif (only present during import; 32% of species), short-term (disappeared quickly; 20%), established-limited spread (survives locally, 30%), and established-widespread (widespread, 18%). Species disappearance rate was highest during ballast deposition and decreased soon after deposition stopped around 1900. Spatial patterns showed a strong association with 19th century railroads for inland dispersal from ports. The disappearance rate and spatial analyses are robust to herbarium collection bias. This study using New Jersey as a model is one of the few documenting multispecies successes and failures in inadvertent plant introductions. Results reveal distinct trends in species establishment and geographic spread and highlight the utility of herbarium specimens in answering questions that span large time scales.