Résumé : From sampling to measurement: a few thoughts about non structural carbohydrates analysis in woody plants.The carbon budget of woody plants has recently become a main goal for plant physiologists. Many studies deal with the way it is affected by climate (temperature or CO
2 levels) or culture conditions (fertilization or water availability). Therefore, the needs for the chemical determination of non structural carbohydrates (NSC) in all the plant organs have strongly increased during the last decade. We intend here to assess the existing methods and analytical procedures currently available with regard to the biochemical and size specificities of woody plants and to the diversity of experimental goals and constraints. Non structural carbohydrates are defined as the sum of starch and sugars soluble in the water present in the cell (hereafter SS). They appear to be very sensitive to degradation by heating or enzymatic transformation, and therefore request important precautions throughout the analytical pathway. Each step of the procedure is detailed and evaluated. The preparation of the sample has to be precisely described since parameters such as sample size, sampling hour and duration, delay before refrigeration affect the final results. Sampling induces a stress which modifies the metabolism of the plant. These perturbations can be limited by cooling, or by freezing (sample immersion) in liquid nitrogen. It must be followed by drying, which is the only way of completely stopping the metabolism of the cell and of avoiding the further degradation of the biochemical compounds. Indeed, some enzymes like invertase are active at ‐‐20°C. The most used, and less damaging, method is freeze‐drying as the combination of vacuum and of low temperature avoids irreversible biochemical changes. Conversely, during hot‐drying, NSC composition can be modified by enzymatic catabolism, starch or sucrose hydrolysis and by browning which occurs above 50°C. Thawing increases those risks since the cell walls have previously been broken by freezing. However, freeze‐drying does not perform a complete water extraction, even when the temperature is increased to 40°C at the end of the process. The accuracy of the chemical determinations depends on the fineness of grinding. For good results in starch analysis, particles have to be smaller than 45 µm. Using a ball grinder which allows the temperature to be kept low by immersing the plant material and bowls in liquid nitrogen is a good method. If the sample could not be dried before grinding, special attention must be given to this last point to avoid thawing and subsequent sample degradation. During storage, the plant material is rehumidified. For instance, peach tree powder, stored during 3 months at ‐ 80°C in airtight pillboxes, took in 1% water. Therefore storage duration before analysis should be limited. Soluble sugar determination is rather sophisticated in woody plants since the different organs exhibit highly variable concentrations and matrix effects. The diversity of the existing methods reflects the diversity of the problems which have to be solved during SS extraction. This first step, leading to supernatant (SS)\residue (starch) separation, is achieved through decantation after soaking, rarely after percolation. Complete SS extraction is generally obtained by repeatedly soaking dried aliquots ( 200 mg) in a volume of solvent which varies from 2 to 35 mL. Accurate and reproducible results may be obtained with only 50 mg of dried aliquot. The extraction medium is sometimes pure water, but it often contains alcohol which avoids biochemical contamination and furthurs the solubilisation of oligosides whose molecular weight is lower than 2,000. Alcohol concentration ranges from 40 to 90% (70‐80%, most usual). Extraction methods also differ by temperature (+ 4°C to 100°C), duration (10 min to 20h) and number (2 to 5) of soaking periods. Next, the supernatant is purified, according to the needs of the chemical determination process. Phenols and pigments are usually eliminated either by chloroform extraction, trapping by polyvinylpyrrolidon (PVP) or animal charcoal. Exchanging ion resin is also used, especially when compounds other than sugars are also determined. Several determination methods are available: enzymatic methods are sensitive and adapted to selective sugar determination and do not request a strong purification of the supernatant. Their use is in constant increase due to their automation and miniaturization; liquid and gas chromatography require a previous preparation of the supernatant. In GC, sugars are transformed into volatile components by derivatization. In HPLC, a severe purification of the supernatant is requested since additional compounds could interfere with chromatogram resolution. These constraints are however compensated by the complete and concomitant quantification of each sugar; colorimetric methods usually allow soluble sugars to be globally determined. But their use tends to decrease in spite of their easy use, complete automation and lower costs. The elimination of phenols in the supernatants before colorimetric resolution is essential. The anthrone method which can be used on alcoholic extracts, is the most currently implemented, in spite of the fact that polyols are not accounted for and although its sensibility depends on the nature of the soluble sugar mixture concerned. Starch determination is usually obtained by glucose determination following starch hydrolysis. Generally, it has to be performed on the residues remaining after soluble sugar extraction even if this step is not indispensable. Starch solubilization is preferably done by autoclaving (1 h, 1 bar, 120°C) as it is softer and more respectful of sample structure than a chemical attack (acid or basic) which could release glucose from the cellulose or hemicellulose chains. For depolymerization, the enzymatic method, which is specific to starch, has to be preferred to an acid or an alcalin hydrolysis. Amyloglucosidase, for which well purified extracts are now available, is often used on its own as a depolymerisation enzyme. The resulting glucose can then be quantified by an enzymatic method. Thus, the results are more accurate than those of colorimetric methods, or those of more expensive and time‐consuming chromatographic methods.