Much attention has been given to recent predictions that widespread extinctions of tropical ectotherms, and tropical forest lizards in particular, will result from anthropogenic climate change. Most of these predictions, however, are based on environmental temperature data measured at a maximum resolution of 1 km2, whereas individuals of most species experience thermal variation on a much finer scale. To address this disconnect, we combined thermal performance curves for five populations of Anolis lizard from the Bay Islands of Honduras with high-resolution temperature distributions generated from physical models. Previous research has suggested that open-habitat species are likely to invade forest habitat and drive forest species to extinction. We test this hypothesis, and compare the vulnerabilities of closely related, but allopatric, forest species. Our data suggest that the open-habitat populations we studied will not invade forest habitat and may actually benefit from predicted warming for many decades. Conversely, one of the forest species we studied should experience reduced activity time as a result of warming, while two others are unlikely to experience a significant decline in performance. Our results suggest that global-scale predictions generated using low-resolution temperature data may overestimate the vulnerability of many tropical ectotherms to climate change.