The leaf intercellular airspace (IAS) is generally considered to have high conductance to CO2 diffusion relative to the liquid phase. While previous studies accounted for leaf-level variation in porosity and mesophyll thickness, they omitted 3D IAS traits that potentially influence IAS conductance (gIAS). Here, we reevaluated the standard equation for gIAS by incorporating tortuosity, lateral path lengthening, and IAS connectivity. We measured and spatially mapped these geometric IAS traits for 19 Bromeliaceae species with Crassulacean acid metabolism (CAM) or C3 photosynthetic pathways using x-ray microcomputed tomography imaging and a novel computational approach. We found substantial variation in porosity (0.04-0.73 m3 m-3), tortuosity (1.09-3.33 m2 m-2), lateral path lengthening (1.12-3.19 m m-1), and IAS connectivity (0.81-0.97 m2 m-2) across all bromeliad leaves. The revised gIAS model predicted significantly lower gIAS in CAM (0.01-0.19 mol m-2 s-1 bar-1) than in C3 (0.41-2.38 mol m-2 s-1 bar-1) plants due to a coordinated decline in these IAS traits. Our reevaluated equation also generally predicted lower gIAS values than the former one. Moreover, we observed high spatial heterogeneity in these IAS geometric traits throughout the mesophyll, especially within CAM leaves. Our data show that IAS traits that better capture the 3D complexity of leaves strongly influence gIAS and that the impact of the IAS on mesophyll conductance should be carefully considered with respect to leaf anatomy. We provide a simple function to estimate tortuosity and lateral path lengthening in the absence of access to imaging tools such as x-ray microcomputed tomography or other novel 3D image-processing techniques.