RATIONALE - Arabusta coffee was created by crossing Coffea arabica L. and C. canephora Pierre ex-Froehner after doubling the chromosome number of C. canephora. For breeding studies and coffee quality improvement, clones have been planted in French Guiana. As green bean metabolite content largely influences the coffee cup quality, biochemical and sensory analyses of Robusta and Arabusta were done on green beans and compared as part of Breedcafs H2020 project. METHODS - Arabusta cherries were harvested on 6 clones in 4 farms in Apatou and in Combi research station, and those from Robusta in Combi research station. They were wet processed before grinding. Phenolic compounds and alkaloids were quantified by HPLC (Campa et al, 2017), sucrose by enzymatic analysis (Megazyme kit, Ireland), diterpenes by NMR (Portaluri et al, 2020) and VOC by headspace SPME-GC/ MS (Marie et al, 2020). Lipids were extracted (Folch et al, 1957) and quantified by cgfb (Bordeaux). Sensory analyses followed the protocol guidelines of the Specialty Coffee Association (SCA) for C. arabica (Marie et al, 2020).We used Statistica software for analyses. RESULTS - Compared to Robusta, Arabusta had a lower caffeine content and higher trigonelline, sucrose and fatty acid contents. Arabusta evidenced both 16-O-methylcafestol and kahweol, known markers for Robusta and Arabica species, respectively. 17 of the 25 non-volatile and 23 of the 47 volatile compounds showed significant differences (p=0.05) between Arabusta and Robusta. Except 5-caffeoylquinate and coumaroylquinate, chlorogenic acids accumulated more in Robusta. A PCA based on volatile compound content separated samples in three groups: Robusta samples, most of the Arabusta from Apatou (richer in pyrazine, toluene, 2-propanone) and those from Combi associated with 3 from Apatou (richer in D-limonene, DMSO, methyl-butanoic acid, and IBMP). A sensory analysis based on 9 sensory parameters indicated that 3 samples received a final score higher than 80