Hesperiidae are claimed to be a group of elusive butterflies that need major effort for sampling, thus being frequently omitted from tropical butterfly surveys. As no studies have associated species richness patterns of butterflies with environmental gradients of high altitudes in Brazil, we surveyed Hesperiidae ensembles in Serra do Mar along elevational transects (900-1,800m above sea level) on three mountains. Transects were sampled 11-12 times on each mountain to evaluate how local species richness is influenced by mountain region, vegetation type, and elevational zones. Patterns were also analyzed for the subfamilies, and after disregarding species that exhibit hilltopping behavior. Species richness was evaluated by the observed richness, Jacknife2 estimator and Chao 1 estimator standardized by sample coverage. Overall, 155 species were collected, but extrapolation algorithms suggest a regional richness of about 220 species. Species richness was far higher in forest than in early successional vegetation or grassland. Richness decreased with elevation, and was higher on Anhangava mountain compared with the two others. Patterns were similar between observed and extrapolated Jacknife2 richness, but vegetation type and mountain richness became altered using sample coverage standardization. Hilltopping species were more easily detected than species that do not show this behavior; however, their inclusion did neither affect estimated richness nor modify the shape of the species accumulation curve. This is the first contribution to systematically study highland butterflies in southern Brazil where all records above 1,200m are altitudinal extensions of the known geographical ranges of skipper species in the region.