Pit membranes in bordered pits between neighbouring vessels play a major role in the entry of air-water menisci from an embolised vessel into a water-filled vessel ( i.e ., air-seeding). Here, we investigate intervessel pit membrane thickness (T PM ) and embolism resistance (P 50 , i.e ., the water potential corresponding to 50% loss of hydraulic conductivity) across a broad range of woody angiosperm species. Data on T PM and double intervessel wall thickness (T VW ) were compiled based on electron and light microscopy. Fresh material that was directly fixated for transmission electron microscopy (TEM) was investigated for 71 species, while non-fresh samples were frozen, stored in alcohol, or air dried prior to TEM preparation for an additional 60 species. T PM and P 50 were based on novel observations and literature. A strong correlation between TPM and P50 was found for measurements based on freshly fixated material ( r = 0.78, P >0.01, n = 37), and between T PM and T VW ( r = 0.79, P >0.01, n = 59), while a slightly weaker relationship occurred between TVW and P50 (r = 0.40, P >0.01, n = 34). However, non-fresh samples showed no correlation between T PM and P 50 , and between T PM and T VW . Intervessel pit membranes in non-fresh samples were c.28% thinner and more electron dense than fresh samples. Our findings demonstrate that T PM measured on freshly fixated material provides one of the strongest wood anatomical correlates of droughtinduced embolism resistance in angiosperms. Assuming that cellulose microfibrils show an equal spatial density, T PM is suggested to affect the length and the shape of intervessel pit membrane pores, but not the actual pore size. Moreover, the shrinking effect observed for T PM after dehydration and frost is associated with an increase in microfibril density and porosity, which may provide a functional explanation for embolism fatigue.