Background/Question/Methods In north temperate conifers, the xylem is optimized to maximize hydraulic efficiency whilst minimizing loss of transport capacity by cavitation. On a 'macro scale', increased cavitation resistance in conifers typically comes at the cost of greater carbon allocation to wood, reduced tracheid diameters and ultimately, reduced transport efficiency. Yet, despite the whole-plant adjustments to cavitation resistance, the actual process of cavitation is thought to occur on the 'micro scale' at the torus-margo pit membrane that connects one tracheid to another. Under functional conditions, the water moves through the porous margo region of the membrane, but should one tracheid become air-filled, the membrane will deflect such that the torus seals the aperture of the functional tracheid and thus prevents air from spreading. Slippage of the torus from its sealing position triggers cavitation. What are the trade-offs that characterize cavitation resistance at the pit level? We sought to answer this question by applying a combination of microscopy (SEM and TEM) to stems belonging to sixteen Cupressaceae species that exhibited a broad range of cavitation resistance. Vulnerability curves were used to assess cavitation resistance, and the 50% loss of conductivity due to cavitation (P50) allowed us to compare the cavitation response across species. Results/Conclusions Our results indicate an increase in torus diameter and a decrease in aperture diameter in response to more negative P50's, a result that is consistent with selection for pits with an improved sealing function at very negative water potentials. However, it is the ratio of the pit aperture to the torus diameter that exhibits the strongest relationship with P50, further supporting the functional interdependence of these features. Decreasing pit aperture caused a nearly 10-fold decrease in aperture conductivity from 34.6 m MPa-1 s-1 in Metasequioa glyptostroboides (P50=-3.76 MPa) to 3.2 m MPa-1 s-1 in Cupressus forbesii (P50=-11.2 MPa) indicating a strong hydraulic trade-off with cavitation resistance. Although tori thickened significantly with increasing cavitation resistance, we detected no clear trend toward altered margo microfibril thickness or margo porosity. Interestingly, basal members of the Cupressaceae phylogeny such as Taxodium and Glyptostrobus have less negative P50's and exhibited weak, poorly differentiated tori with reduced margo complexity as compared to more derived, resistant taxa such as Sequoiadendron and Juniperus. Our data suggest that radiation of the Cupressaceae into drier habitats selected for increasingly specialized torus-margo pit membranes.