Introduction – Nuclear magnetic resonance (NMR) is one of the most commonly used analytical techniques in plant metabolomics. Although this technique is very reproducible and simple to implement, sample preparation procedures have a great impact on the quality of the metabolomics data. Objective – Investigation of different sample preparation methods and establishment of an optimised protocol for untargeted NMR-based metabolomics of Vitis vinifera L. wood samples. Methods – Wood samples from two different cultivars of V. vinifera with well-defined phenotypes (Gamaret and 2091) were selected as reference materials. Different extraction solvents (successively, dichloromethane, methanol and water, as well as ethyl acetate and 7:3 methanol-water (v/v)) and deuterated solvents (methanol-d4, 7:3 chloroform-d-methanol-d4 (v/v), dimethylsulphoxide-d6 and 9:1 dimethylsulphoxide-d6-water-d2 (v/v)) were evaluated for NMR acquisition, and the spectral quality was compared. The optimal extract concentration, chemical shift stability and peak area repeatability were also investigated. Results – Ethyl acetate was found to be the most satisfactory solvent for the extraction of all representative chemical classes of secondary metabolites in V. vinifera wood. The optimal concentration of dried extract was 10 mg/mL and 7:3 chloroformd-methanol-d4 (v/v) was the most suitable solvent system for NMR analysis. Multivariate data analysis was used to estimate the biological variation and clustering between different cultivars. Conclusion – Close attention should be paid to all required procedures before NMR analysis, especially to the selection of an extraction solvent and a deuterated solvent system to perform an extensive metabolomic survey of the specific matrix.