All protocols for Next Generation Sequencers (NGS) start with the preparation of a DNA library from extracted DNA. Quality, number, and length of the sequences produced by NGS all depend on the quality of the library that was prepared, itself dependent on the DNA quality. Generally, when using NGS, one would wish to obtain the genomic information contained in the nucleus without an excessive proportion of cytoplasmic DNA. Multiple copies of cytoplasmic genomes are present in every cell (11 to 70 copies, depending on the developmental stage and/or physiological status) ( Tymms et al., 1983 ). The chloroplast DNA of plants, therefore, represents 17 - 23% of their total DNA ( Boffey and Leech, 1982 ). In this study, we developed a protocol that strongly reduces the amount of cytoplasmic DNA. The improvement in quality makes the method perfectly adapted to library construction for NGS. We applied this extraction method to the grapevine ( Vitis vinifera L.) and coffee tree ( Coffea canephora Pierre ex A. Froehner). These species are known to contain many secondary metabolites, which can generate diffi culties for DNA extraction ( Peterson et al., 1997 ; Mattivi et al., 2006 ). To establish the protocol, we were partially inspired by a classic protocol for nuclear DNA plant extraction destined for the construction of BAC (Bacterial Artifi cial Chromosome) libraries ( Peterson et al., 1997 ). The two stages of the protocol are: (i) isolation of nuclei and (ii) nuclear DNA extraction. Nuclear DNA obtained with this protocol was then sequenced using 454/GS-FLX technology (Roche, Basel, Switzerland) with the Titanium kit. All protocols for Next Generation Sequencers (NGS) start with the preparation of a DNA library from extracted DNA. Quality, number, and length of the sequences produced by NGS all depend on the quality of the library that was prepared, itself dependent on the DNA quality. Generally, when using NGS, one would wish to obtain the genomic information contained i