A new method of Random Square-quadrats Method (RSQM) is used to characterize the spatial patterns and spatial associations of five dominant tree species in a Korean pine broadleaved old-growth forest in Changbai Mountains in this paper. RSQM is developed based on the formula ICS(t)=S2(t)(t)-1, and can measure the spatial distribution patterns at a range of spatial scales t. The confidence intervals of 95% and 99% are constructed using random simulation technique to test the significance of any deviations from the null hypothesis of complete spatial randomness or type independence. Results show that, Quercus mongolica, Malus baccata and Maackia amurensis are randomly distributed in the study plot, and the other tree species are significantly aggregated at local scales. Most of species pairs and species groups exhibit significantly positive spatial associations at local scales. The results demonstrate that RSQM differs from the traditional small plot method as well as from point pattern analysis method (Riply′s K). The traditional small plot method measures the spatial pattern using aggregated indices of dispersion, but it can only characterize the spatial pattern at a given scale. Point pattern analysis method based on distance of points is very popular but susceptible to edge effects since one can not count points outside of the study plot. RSQM can avoid these disadvantages. It can describe the distribution patterns at a set of distances and isn′t influenced by edge effect. Especially, it is very effective in describing the spatial relathionship whitin multi-species.