Abstract: The plant population pattern is a real charocteristic in a forest community. Like all statistical methods used in the study of the phytocoenology, the results of the testing and analysis of the plant population pattern depend heavily .on the field sampling method used with the community. In three different communities in the Ding Hu Shah natural reserve, Guangdong, China, the following five methods were applied to the sampling for information on plant species distribution: 1. Random quadrats, 2. contiguous grid quadrats (Greig-Smith, 1952;Kershaw, 1957), 3. point-centered quarter(Bray, 1962), 4. nearest neighour(Clark and Evans, 1954) and 5. closest individual (Pielou, 1959).The comparative studies dealing with the arrangement of the field data, the pattern testing and the pattern analysis showed that method 1. was affected by the quadrat size, although it can be used with all the communities. The use of the plotless methods is limited in the evergreen broad-leaved forest. The method 2. was an effective one with every community not only for pattern testing but also for pattern analysis. The various methods gave different testing results with one and the same in a community, while each method presents a difference in adaptation to different communities.The results of the research suggests that method 3. is insensitive to the departure from random dispersion, method 4. and 5. are suitable to the mono-dominant community, and the contiguous grid quadrats method is a multi-purpose approach that can be used widely for the research of plant population pattern in a subtropical forest community.