This paper describes an approach of tourism development from tourists’ perspective by analyzing tourist review data. The approach enables us to understand characteristics of regions and tourists based on sightseeing themes extracted by applying text mining and PLSA to tourist review data. This study extracted 15 themes from about 170 thousand review data on tourist spots across Japan; scenery, historical buildings, sea views, shopping areas, flowers, shrines and so on. A regional tourism marketer can quantitatively understand characteristic themes of the region and target tourists who show interest in these themes. Some regions with similar themes can get a round trip effect by cooperating to establish a tour route experiencing the theme. Some regions with different themes can get new targets by cooperating to mutually send tourists showing interests in respective themes. The approach will be useful for tourism planning and regional developments to attract and pull in more tourists.