Resumé : The 4TREE Axiom 50K SNP array is one of the main deliverables of the H2020 project B4EST which aims at transferring know-how and tools to improve breeding for production, resilience and diversity of forest trees. This array is developed for: Populus sp, Fraxinus sp, Pinus pinaster and Pinus pinea. It is based on de novo sequencing resources and existing genomic resources of loci of interest for breeders. 4TREE is primarily aimed at breeders by providing low genotyping costs for economically important species in Europe, fostering collaboration between breeding programs across Europe. One of its first uses will be genome wide association studies and genomic predictions, being also extremely useful for monitoring and characterizing novel sources of variation. Material/ Methods : The 4Tree array will be designed after the screening of 420K SNP. Except for the orphan species P. pinea, SNP were identified from existing genomic resources: whole genome and RNA sequencing, exome capture, SNP arrays. SNP gathered by these resources were filtered for inclusion in a screening array of 420K (105K/species) and checked over a panel of 480 samples (120/species). SNP without other polymorphisms within 30bp flanking sequences were preferentially selected and those in unique region were prioritized, limiting A/T and C/G which require more space on the array. The selection included several SNP in candidate genes or regions found via transcriptomic, QTL and GWAS analysis. Results : For P. pinea, 128K SNPs were submitted to ThermoFisher, resulting in 37K recommended SNPs for the screening array. P. pinaster submitted 197K SNPs (5K SNPs from Plomion et al. 2016), leading to 140K inclusions in the screening array, including several from candidate genes for abiotic and biotic stress. For Fraxinus sp, 126K SNP were selected from the 327K scored. SNPs were selected in order to prioritize candidate SNPs associated to dieback disease tolerance and incompatibility (Stocks & Metheringham et al. 2019). SNPs were additionally selected according to the genomic coverage (maximum possible representation of genes and scaffolds with respect to the reference genome). For Populus, a subset of SNPs was identified for two species: P. nigra and P. deltoides. These SNPs were either species specific or shared between the two species. For P. nigra, SNPs from a previous array (Faivre-Rampant et al. 2016) and from candidate regions from GWAS were also selected. The final coverage of the genome was 100 SNP/Mb with a ratio of genic and intergenic SNPs of 60/40%. In total, 3 000K SNP were submitted to ThermoFisher for Populus to select 120,500 for the screening array. Conclusion: From this screening genotyping, 12.5K high value SNPs per species will be retained for the final array 4TREE dedicated to gene discovery, pre-breeding monitoring of diversity and genomic selection studies in B4EST. This final array will be a highly valuable genome-assisted breeding tool, made commercially available for cost effective HT genotyping for four important tree species in Europe.