Plant species that occur in temperate areas in North and South America, but are absent from interleaving tropical areas, is one of the most striking biogeographic patterns.  This distribution pattern, referred to as the American amphitropical distribution, has evolved in 237 land plant species.  Despite this relatively common distribution, we know little about the biogeographic processes that have led to such a remarkable pattern.  Three hypotheses have been postulated to explain the American amphitropical distribution pattern.  The first is long-distance dispersal.  Some evidence has supported long-distance dispersals from North to South America being the predominant pathway.  The second is vicariance, in which species historically occupied temperate and tropical areas continuously, but interleaving populations have since gone extinct.  The third is a stepping stone model in which smaller patches of suitable habitat intermittently connect temperate areas and allowed for multiple migration and establishment events that occurred repeatedly in relatively shorter distances.  To test the above hypotheses, we applied a species distribution modeling approach to estimate the probability of species occurrences based on current distributions and the species' bioclimate conditions.  Under the vicariance model, we predicted that a continuous distribution of suitable environmental conditions would connect the Northern and Southern Hemispheres, even if species do not currently occupy those areas.  Under the stepping stone hypothesis, we predicted that discontinuous patches of suitable habitat would connect the Northern and Southern Hemispheres.  Although our approach does not allow us to explicitly test for long distance dispersal, rejecting the stepping stone and vicariance hypotheses would inform the probability of long distance dispersal as the remaining hypothesis.  We tested our hypotheses with 25 species based on current bioclimate data and distributions, and projected the models into current climates and the last glacial and interglacial maxima.  The majority of species distribution models across time scales failed to show suitable tropical habitat based on the model's thresholds per species.  We rejected the vicariance hypothesis in all but one case (Tiqullia nuttallii [Boraginaceae] in the last interglacial maximum).  We failed to reject the stepping stone hypothesis in 11 out of 75 analyses.  Although the vicariance and stepping stone models remain a possible explanation for the amphitropical pattern, our best explanation is that long distance dispersal was the predominant mode.