The reliability of a double-ended pressure sleeve technique was evaluated on three woody angiosperm species with contrasting maximum vessel lengths. Vulnerability curves (VCs) were constructed by varying sample length and the size of the pressure sleeves. VCs were compared against curves obtained with reference techniques. For the two diffuse-porous species, Betula pendula and Prunus persica, VCs built with shoot segments shorter than maximum vessel length strongly overestimated species vulnerability. Furthermore, increasing the size of the pressure sleeve also tended to lead to overestimated VCs. For the ring-porous species Quercus robur, the technique strongly overestimated vulnerability to embolism, whatever the sample length or chamber tested. In conclusion, the double-ended pressure sleeve technique only gives reliable VCs on diffuse-porous angiosperms with short pressure sleeves, only when segments are longer than maximum vessel length.