Ever since MacArthur and Wilson presented the Theory of Island Biogeography in 1963, there has been considerable concern about the ability of a system of isolated reserves to retain species diversity or viable populations over extended periods of time. Despite this interest, very little is known about how readily individuals move around in their landscape. Edges and corridors have both been targeted as landscape features that may have important consequences for directing movement and thereby affecting isolation. Edges may form a barrier to movement reducing the probability that an individual will leave a patch. For individuals that have left, they must find a new patch, and corridors may help direct them there. We studied the responses of butterflies to these two important landscape features.;We tracked the responses of two butterflies: the regal fritillary (Speyeria idalia) and the monarch (Danaus plexippus) at four edge types (crop, field, road, treeline) in the highly fragmented prairies of central Iowa. S. idalia responded strongly to all except road edges, but their responses were strongly influenced by conspecific density. Individuals in high density areas were less likely to leave the prairie. Surprisingly, S. idalia showed a very strong response to edges with very subtle differences in vegetation structure, such as a prairie and a pasture. D. plexippus responded only to treeline edges. Their responses were more influenced by factors such as wind, nectar availability, and time of year. We also explored the conservation value of roadside vegetation, both as additional habitat and corridors or stepping stones between prairies.;We examined the abundances of common, open area butterflies and the species richness of grassland specialists in three types of roadsides: grassy, weedy, and restored to prairie. We also considered how distance from a source prairie affected patterns. Two common butterflies, D. plexippus and Everes comyntas showed a significant relationship with roadside type, but not distance. There were no significant associations between grassland species richness and any factor we measured. More intensive sampling may be necessary, though, to detect an effect.