Detecting range shifts and contractions is critical for determining the conservation priority of rare and declining taxa. However, data on rare species occurrences frequently lack precise information on locations and habitats and may present a biased picture of biogeographic distributions and presumed habitat preferences. Herbarium or museum specimen data, which otherwise could be useful proxies for detecting temporal trends and spatial patterns in species distributions, pose particular challenges. Using data from herbaria and Natural Heritage Programs on numbers of occurrences within individual municipalities (towns, cities, or townships), we quantified temporal changes in the estimated distributions of 110 rare plant species in the six New England (USA) states. We used the partial Solow equation and a nonparametric test to estimate the probability of observing multiple absences (gaps in the collection record) if a given population was actually still extant. Bayes' Theorem was used to estimate the probability that occurrences were misclassified as extinct. Using the probabilities obtained from these three methods, we eliminated taxa with high probabilities of pseudo-absence (that would yield an inaccurate profile of species distributions), narrowing the set for final analysis to 71 taxa. We then expressed occurrences as centroids of town polygons and estimated current and historical range areas (extents of occurrence as defined by alpha-hulls inscribing occurrences), mean distances between occurrences, and latitudinal and longitudinal range boundaries. Using a geographic information system, we modeled first, second, and third circular standard deviational polygons around the mean center of the historical range. Examining the distribution of current occurrences within each standard deviational polygon, we asked whether ranges were collapsing to a center, expanding, fragmenting, or contracting to a margin of the former range. Extant ranges of the species were, on average, almost 67% smaller than their historical ranges, and distances among occurrences decreased. Five New England hotspots were observed to contain >35% of rare plant populations. Extant occurrences were more frequently marginalized at the periphery of the historical range than would be expected by chance. Coarse-grained data on current and historical occurrences can be used to examine large suites of species to prioritize taxa and sites for conservation.