Southern forests are at the highest ozone (O3) risk in Europe where ground-level O3 is a pressing sanitary problem for ecosystem health. Exposure-based standards for protecting vegetation are not representative of actual field conditions. A biologically-sound stomatal flux-based standard has been proposed, although critical levels for protection still need to be validated. This innovative epidemiological assessment of forest responses to O3 was carried out in 54 plots in Southeastern France and Northwestern Italy in 2012 and 2013. Three O3 indices, namely the accumulated exposure AOT40, and the accumulated stomatal flux with and without an hourly threshold of uptake (POD1 and POD0) were compared. Stomatal O3 fluxes were modeled (DO3SE) and correlated to measured forest-response indicators, i.e. crown defoliation, crown discoloration and visible foliar O3 injury. Soil water content, a key variable affecting the severity of visible foliar O3 injury, was included in DO3SE. Based on flux-effect relationships, we developed species-specific flux-based critical levels (CLef) for forest protection against visible O3 injury. For O3 sensitive conifers, CLef of 19mmolm-2 for Pinus cembra (high O3 sensitivity) and 32mmolm-2 for Pinus halepensis (moderate O3 sensitivity) were calculated. For broadleaved species, we obtained a CLef of 25mmolm-2 for Fagus sylvatica (moderate O3 sensitivity) and of 19mmolm-2 for Fraxinus excelsior (high O3 sensitivity). We showed that an assessment based on PODY and on real plant symptoms is more appropriated than the concentration-based method. Indeed, POD0 was better correlated with visible foliar O3 injury than AOT40, whereas AOT40 was better correlated with crown discoloration and defoliation (aspecific indicators). To avoid an underestimation of the real O3 uptake, we recommend the use of POD0 calculated for hours with a non-null global radiation over the 24-h O3 accumulation window. © 2015 Elsevier B.V.