Abstract：In this study we chose the Taroko National Park as the research site, which is a popular sightseeing park with the characteristics of the natural resources and historic culture. First, the exploratory factor analysis was applied to extract five cognitive, affective, and unique factors of the destination images for the tourists, who visited Taroko National Park. Second, the cluster analysis was employed to identity four clusters among different factors of the destination images. Finally, the travel cost method was adopted to establish Taroko National Park recreation demand model, in order to estimate the recreation demand function by On-Site Poisson estimation model and to measure the differences in recreation benefits of the four destination image clusters. The empirical results of this study are as follows: (1) Among four identified clusters, the multi-purpose image seekers were found to be the most important segment. (2) The four destination image clusters differ significantly in terms of national park demands. (3) The recreation benefits of visiting Taroko National Park ranged between NT $3,684 to 4,139 for every visitor per year. (4) The “landscape attraction clusters” had the higher recreation benefits than other three destination image clusters. This study concluded that administrators of Taroko National Park should pay more attention to the destination image and recreation demand of domestic and international visitors in order to achieve the goals of sustainable development for the national park. Furthermore, attracting primary clusters by way of market segmentation, and the resource maintenance program to protect the natural landscape, ecological resources and recreation quality need to setup. The operating profit gained from recreation and environmental resources management through the entry fees could be invested in the management of Taroko National Park to improve the primary facilities and services, and enhance the ecological quality. (Full text)