Trampling usually causes a serious disturbance on vegetation, and its measurement and control influence the management policy of a recreation area. Two vegetation types (intolerant and tolerant) of grassland in Heng-Chun Tropical Botanical Garden were examined by experimental trampling in early summer. Trampling treatments ranged from 0 to 1,600 passes along trampling lanes. Changes in species cover, species persistence, mineral soil exposure and soil penetration resistance were investigated. After the treatment, the mean covers of two vegetation types decreased as trampling intensity increased, rapidly at low levels of trampling but slowly at the intensity over 800 passes. This asymptotic relationship indicates that trampling damages can be minimized by concentrating use in space instead of dispersing over large areas. In contrast to vegetation loss, bare soil was exposed after trampling intensities over 200 passes, and the exposure area increased linearly with trampling intensity. Therefore, wherever use intensities are high it is better to consider a site that is resistant to exposure of mineral soil than to select a site with resistant vegetation. Both species persistence, representing the changes of species, and soil penetration resistance, which affected by vegetation types and trampling intensity, changed slightly as trampling intensity increased.