The distribution pattern of plant species in communities is closely related to spatial scale. Quadrat sampling and traditional methods only can analyse pattern in one scale. Therefore, it is difficult to reflect species spatial characteristics completely, no matter how rational for quadrat size. Point pattern analysis, based on species spatial mapped points, is introduced in this paper. Three advantages over traditional techniques, emerged in the analysis of three dominant species in oak forests in Clinton county Michigan, USA. Firstly, it can analyse species spatial patterns and interspecific associations at any scale. Secondly, its results are more consistent to the reality, especially for community structure. Thirdly, it is more suitable for study of relationships between species pattern and time scale. The three species analysed in this paper have obvious clumping feature, while they vary with scales.