Coffee is one of the most important agricultural commodites in the world. Coffea arabica, an allotetraploid from a recent hybridization of two diploid species (C.canephora and C.eugenioides) is responsible for 70% of world commercial production. Recent studies have been done to analyse transcriptome data of Coffea spp. in order to improve our knowledge in genetics and gene expression of those species. However, there is few data on coffee fruit transcriptome. In this work, RNA-Seq of two organs (whole fruit - cherry stage and mature leaves) from allotetraploid C. arabica cv. Mundo Novo and its diploid ancestor C.eugenioides, were done using Illumina HiSeq 2000 (100 bp single-end sequences). We report the generation of near 10 million reads for C.arabica (4,747,049 of fruits and 6,156,750 reads of leaves) and 7,5 million reads for C.eugenioides (3,688,364 of fruits and 3,835,373 reads of leaves). De novo assembly and digital gene expression analyses was performed to generate a collection of 35,462 contigs from C.arabica cv. Mundo Novo (average length of 691 bp) and 36,935 contigs of C.eugenioides (average length of 701 bp). Funnctional annotation were performed by sequence comparison with public databases (NCBI-nr, Swiss-prot and PlantCyc). Our results showed the presence of prevalent genes and species-specific either in C.arabica or C.eugenioides that could explain particular characteristics of these two species. Characterization of Coffea spp. transcriptome provides an effective tool for better understanding for differential gene expression in Coffea spp., providing important clues of C.arabica gene evolution, as well as valuable information for coffee breeding.