Two experiments, one comprising 12 Eucalyptus clones at three different spacings (2 x 2 m (2,500 trees ha~'), 3 x 3 m x 1111 trees ha) and 4 x 4 m (625 trees ha~ ) and the other comprising 12 Eucalyptus full-sib familles at two spacings (2 x 2 m and 4 x 4 m) were established to study the effect of spacing on heritability. Volume was measured on a monthly basis up to 30 months after planting. As the clone or family sample was small, the fixed model was used, and the ratio (F-l)/F (F for Fischer) was used as an estimator of the heritability calculated on genotype mean basis. Results differed according to genotype. For clones, the differences of heritability between all three spacings are very small (<.2). The (F-l)/F ratio for 625 stem ha~' was higher because of a weaker environmental effect at this spacing. In conclusion, the clone experiment showed that the increased planting density should have little effect on the efficacies of clone selection in clonal tests. However, for familles, the heritability was higher for 2,500 stem ha~' because of a higher genetic variance. Differences in the heritability estimates between the 2 spacings were more than 0.4, but tended to decrease with age. In conclusion, the family experiment showed that the increased planting density should improve the efficiency selection. Differences in heritability trend with age between the 2 experiments are also discussed. With large square plots, the structure of genotype, i.e., the presence or absence of within genotype variance, appears to be the best explanatory factor.