An Open-Top Chambers experiment on Fagus sylvatica and Quercus robur seedlings was conducted in order to compare the performance of an exposure-based (AOT40) and a flux-based approaches in predicting the appearance of ozone visible injuries on leaves. Three different ozone treatments (charcoal-filtered; non-filtered; and open plots) and two soil moisture treatments (watered and non-watered plots) were performed. A Jarvisian stomatal conductance model was drawn up and parameterised for both species and typical South Alpine environmental conditions, thus allowing the calculation of ozone stomatal fluxes for every treatment. A critical ozone flux level for the onset of leaf visible injury in beech was clearly identified between 32.6 and 33.6 mmolO3 m(-2). In contrast, it was not possible to identify an exposure critical level using the AOT40 index. Water stress delayed the onset of the leaf visible injuries, but the flux-based approach was able to take it into account accurately.