An international project aiming to sequence a large collection of 180,000 cocoa cDNA enriched in full length cDNAs was carried out. CDNA were isolated from different organs, sometimes submitted to various biotic and abiotic stresses. About 45 cDNA libraries were constructed from a wide panel of organs: flowers, flower cushions, seeds at different stages of development and during fermentation, cherels, pod cortex, shoot, wood, root, germinated seeds, and embryos from vitro culture. Direct and SSH libraries were constructed from organs subjected to biotic or abiotic stresses: leaves and pods inoculated with Phytophthora palmivora and Phytophthora megakarya, shoots and/or pods inoculated with Crinipellis perniciosa, Moniliophthora roreri, Ceratocystis fimbriata, and shoots attacked by mirids. leaves and roots under drought conditions. The construction and management of cDNA libraries, was performed with the help of the robotic platform, part of the "Montpellier Languedoc-Roussillon GENOPOLE". All the sequencing work was performed by GENOSCOPE (from the GENOPOLE Evry/France). A bioinformatic pipeline (ESTtik) was constructed; it automatically processes the sequences, assembles, annotates them, and integrates the results into a web-based database, allowing researchers browse and query of the results. About 62% of cocoa sequences showed significant similarity to gene sequences from other species and a part of them could be classified according to the Gene Ontology. This new molecular cocoa resource allowed the identification of hundreds of new microsatellite markers and thousands of SNPs (Single Nucleotide Polymorphisms). It will allow to identify an unigene set suitable for all functional genomic studies in cocoa. (Texte intégral)