Spatial disribution was analysed for all plot trees and representative tree species in two duplicate l-ha plos established in a foothill rain forest 16 km west from Padang, West Sumatra, Indonesia.All nee trunks ) 8 cm dbh were distributed randomly in both plots. Spatial conelation between dead trees and recruited trees during the 8-year period was locally negative while broadly positive, suggesting that the random spatial pattern is dynamicatly maintained by mechanisms of density-dependent regulation.All of five abundant canopy/subcanorpy non-pioneer species, i.e. Hopea dryobalanoi.des, Gonystylus forbesii, Cleistanthus glandulosus, Mastixia trichotoma, and Grewia florida, and typical pioneer Macaranga spp., including M. gigantea, M. hypoleuca, M. pruinosa and M. triloba, showed a clumped distribution, particularly for smaller size classes. Two subcanopy non-pioneer species, Mastixia and Grewia showed a shift from a regularly decreasing density with height in one plot with larger gap area to bimodal height distribution in the other plot with smaller gap area. They also showed a similar tendency in spatial pattern such that the negative spatial correlation was observed between seedlings (< I m high) and poles (1-10 m high) and bemeen poles and tall rees (> 10 m high) in the plot with larger gap area but that was found between seedlings and tall trees in the plot with smaller gap area. Such pattern was also observed for gapdemandrng Macaranga spp., and they were concluded to be less tolerant than Hopea, Gonystylus and Cleistanthus, in which no height distibution difference nor spatial patt€rn difference was observed between two plots. These more tolerant species showed a positive correlation between tall trees and seedlings, but no correlation between tall fees and poles. These The census for seedlings and saplings was carried out by tagging numbered tapes, measuring of top height and dbh (when taller than 1.3 m), and mapping the position of the stem base. We also measured height and dbh of all nees of these species ) 8 cm dbh. We recorded coordinates of all individuals from dispersion maps of two plots using a digitizer, and devised a program to draw maps of any species and/or categories using a microcomputer.The dispersion maps of each species are shown in Appendices l-6. We applied the statistic indices of spatial distribution and correlarion of lwao (1977). T11e m*/m index (the mean-crowding divided by mean density) of spatial distribution of a given populationislforarandomdistribution,>lforaclumpeddistribution,and