The interaction assessment of genotype and environment is necessary to find out an effective selection strategy in progeny trial, particularly in advanced generations. This study aims to observe the effect of family × site interaction in second generation progeny trial of Eucalyptus pellita planted at two locations: Wonogiri, Central Java and Pelaihari, South Kalimantan. The trials design were completely randomized block design (CRBD) with 49 families, 6-10 blocks, 5 tree-plot and 4 m × 1.5 m spacing. Data measurements included height, diameter at breast height, and stem volume at two and four years age. The study results showed that the families had significant differences (p<0.01) for all measured traits at two and four years age, but the family and site interaction was not statistically significant. Family heritability across the two sites were moderate to high for all traits with the type B genetic correlation ranged from 0.67 to 0.94. Heritability and genetic correlation tended to be higher as tree getting older. Less than 16% of families showed an interactive rank changes in the two locations based on multiple-traits selection index. In general, family selection by combining data from the two sites of progeny trial provided higher genetic gain than those by indirect selection, but it was varied to those direct selection on each site. Therefore, these results imply that, the strategy for family selection by combining data from the two sites of progeny trials should be practiced in appropriate combination and direction by considering the magnitude of genetic variation.