The understanding of the genetic bases of coffee bean chemical composition is a requirement for breeding programs, which aim to improve coffee beverage quality. Nowadays, the development of the genomic toolkit allows the identification of candidate genes potentially controlling these traits. However, the direct utilization of these new tools for breeding relies on the ability to identify within this set of candidates the ones that are responsible for the variability observed between genotypes, i.e. the ones whose polymorphisms affect the variability of the traits of interest. Evaluation of the nucleotide diversity of the genes, which is a pre-requisite to test these associations, can be done in two ways: through the analysis of the EST libraries available (in silico strategy) or through direct sequencing of genotypes of interest (in vivo strategy). The purpose of this study was to evaluate the relevance of the in silico polymorphism discovery strategy in Coffea based on the EST libraries currently available. Genes for the biosynthetic pathways leading to sucrose and coffee specific diterpens (cafestol and kawheol) were analysed in silico. This strategy yielded the identification of 1.1 polymorphism per 100 bp in average for the 14 analysed genes, this result underlying the feasibility of this method for Coffee. Analysis of the nucleotide diversity in vivo for a few genes of the sucrose biosynthesis pathway allowed a comparison of the two strategies. The in silico discovery strategy is complementary to the in vivo one providing in a cost effective manner a first evaluation of nucleotide diversity at the whole genome level.