Heterozygote deficiencies in natural populations of outbreeding tree species are common and thought to be due mainly to biparental inbreeding. Inbreeding is believed to be caused by family structure within populations, a product of limited seed dispersal and probably limited pollen dispersal. Although both theory and simulation studies predict that structure should be apparent where trees are isolated by distance, most studies of structure in natural populations have detected only a weak spatial genetic structuring. In this contribution, we compare the use of spatial autocorrelation methodology and F statistics with the concept of relatedness to examine the spatial genetic structure in the natural population of a native southern beech and to explore the discrepancy between theory and observations. Autocorrelation detected structure in only a few of the nine enzyme loci tested in an estimated patch size of approximately 10 m. By successively eliminating the largest distances in the Gabriel map, the population was separated into groups or patches of neighbors, which were then tested for relatedness. Three groups of relatives were found interspersed with seven groups of unrelated individuals. The F statistics for these groups also showed weak genetic structure. We suggest that heterogeneity of family structure within natural populations may be one reason why more spatial genetic structure has not been detected