Abstract: The south-subtropical evergreen broad-leaf forest of Mt. Dinghu has been studied by using both the "quadrat method" and "point-centred quarter method". Twenty 10 × 10m. quadrats and fourty point-centred quarter samples were restricted at random (partial random) and arranged in the same stand. Minimal area of the quadrat samples of the arborescent strata is 1000–1200m2 and that of the undergrowth is eight to ten 2Ï2m. small quadrats. It is indicated by species-area curves (Fig. 1, 2, 4). We suggest a concept of "minimal point number", which is determined by the correlation curve between species and point-centred quarter number. This correlation curve is termed "species-point number curve" (Fig. 3.), which indicated minimal point number of the south-subtropical evergreen broad-leaf forest is 20 to 22 point-centred quarter samples. It has been shown that the point-centred quarter method has higher precision and higher efficiency than the quadrat samples as judged by importance value, percentage of similarity and coefficient of community (Table 1, 2, 3). So that, the point-centred quarter method is suitable for the south-subtropical evergreen broad-leaf forest used as a sampling technique.