Species ranges are shaped by both climatic factors and interactions with other species. The stress gradient hypothesis predicts that under physiologically stressful environmental conditions abiotic factors shape range edges while in less stressful environments negative biotic interactions are more important. Butterflies provide a suitable system to test this hypothesis since larvae of most species depend on biotic interactions with a specific set of host plants, which in turn can shape patterns of occurrence and distribution. Here we modelled the distribution of 92 butterfly and 136 host plant species with three different modelling algorithms, using distribution data from the Swiss biodiversity monitoring scheme at a 1 × 1 km spatial resolution. By comparing the ensemble prediction for each butterfly species and the corresponding host plant(s), we assessed potential constraints imposed by host plant availability on distribution of butterflies at their distributional limits along the main environmental gradient, which closely parallels an elevational gradient. Our results indicate that host limitation does not play a role at the lower limit. At the upper limit 50% of butterfly species have a higher elevational limit than their primary host plant, and 33% have upper elevational limits that exceed the limits of both primary and secondary hosts. We conclude that host plant limitation was not relevant to butterfly distributional limits in less stressful environments and that distributions are more likely limited by climate, land use or antagonistic biotic interactions. Obligatory dependency of butterflies on their host plants, however, seems to represent an important limiting factor for the distribution of some species towards the cold, upper end of the environmental gradient, suggesting that biotic factors can shape ranges in stressful environments. Thus, predictions by the stress gradient hypothesis were not always applicable.