Dendroecological studies have clearly pointed out delayed effects of climatic stresses on radial growth, especially drought: tree ring width is reduced the year of drought, but the fol-lowing tree ring is often narrower than the climate would predict. Moreover, we already dem-onstrated that stand leaf area index is reduced one or several years after a severe drought. Nevertheless, the underlying physiological processes are up to now not demonstrated. One may hypothesize that reduced photosynthesis due to stomatal closure will first stopped tree growth (direct effect) and secondly limited carbohydrates storage. This reduced storage would impair the following spring through reduced foliation, partial mortality and limited growth. The aim of the presented study is to analyse the effect of drought 2003 on the replenish-ment of total non-structural carbohydrates amount in oak and beech trees. In a 35-40 year old beech stand, we selected 35 couples of dominant trees with premature (end of August) or normal leaf fall (end of October). We cored trees to analyse stem TNC content at the end of the growing season (October 2003). The premature defoliated beech exhibited significant lower starch content than normal defoliated tress, but no dead tree was observed in the 2004 spring. Phenological observations in Spring 2004 showed twigs mortality, heavy fructi-fication, small leaves and we will discussed links between theses events and the 2003 physiological status (defoliated or not, TNC content). The Harth forest (Alsacian Plain) is an oak forest highly responsive to soil water shortage as a result of low soil holding capacity, high proportion of stones and high stand leaf area index. The recent oak decline has been retrospectively ascribed to recurrent droughts. In this forest, only light premature leaf-fall was observed in 2003 but brown and burned leaves were noted. More than 450 sessile or pedunculate oaks were there cored in October 2003, per-cent of brown leaves and leaf fall evaluated and TNC content quantified. We computed sev-eral water stress indices, taking into account local soil properties, stand leaf area index and species composition, as well as climatic data. We analysed the within and between stands variability of TNC content, especially starch, with respect to Spring 2004 foliation, twigs mor-tality and stand water balance. The relationships between drought intensity and phenologi-cal responses or carbohydrates reserves status of the two oak species will be discussed.