ABSTRACTRelationships between hydraulic vulnerability expressed asP50(the air pressure causing 50% loss of hydraulic conductivity) and within-ring differences in wood density (WD) and anatomical features were investigated with the aim to find efficient proxies forP50relating to functional aspects. WD and tracheid dimensions were measured with SilviScan on Norway spruce (Picea abies(L.) Karst.) trunk wood.P50was strongly related to mean WD (r= -0.64) and conduit wall reinforcement ((t/b)2), the square of the ratio between the tracheid double wall thickness (t) and the lumen width (b), where use of tangential lumen width ((t/bt)2) gave better results (r= -0.54) than radial lumen width (r= -0.31). The correlations ofP50with earlywood (EW), transition wood (TW) and latewood (LW) traits were lower than with the specimen averages, both for WD (r= -0.60 for WDEW,r= -0.56 for WDTW,r= -0.23 for WDLW) and all anatomical traits. The loss of hydraulic conductivity was addressed as a dynamic process and was simulated by defining consecutive phases of 5% theoretical conductivity loss. WD and tracheid traits were calculated and correlated withP50values of each specimen. Tightest correlations were found for (t/bt)2, at relative cumulated theoretical conductivities until 45 to 50% (r= -0.75).We conclude that WD is one of the best available proxies forP50, but does not necessarily reflect the mechanism behind resistance to cavitation. The new trait, based on estimation of conductivity loss as a dynamic process, provided even stronger correlations.