Biological sciences in general have benefitted from rapid technology developments producing an ever-growing number of large, multidimensional data sets. In the context of structural imaging, this includes the move from 2D section-based insight to true 3D data collection. One of the most useful techniques allowing for exploration of cellular structures in 3D with unparalleled fidelity and nanometre resolution is electron tomography (ET). In highly compartmentalised cells, such as cardiomyocytes, elaborate sub-cellular membrane structures play crucial roles in cellular electrophysiology and mechanics. Although the anatomy of specific ultra-structures, such as dyadic couplons, has been described using 2D electron microscopy (EM) of thin sections, we lack accurate quantitative knowledge of fine 3D spatial interrelations of other sub-cellular components. Here we illustrate the utility of 3D ET for identification, visualisation, and analysis of cardiac ultrastructures such as T-tubular system (T-tub). Ventricular T-tub form an intricate system of surface membrane invaginations that, through their close connections with SR, support spatially and temporally synchronous excitation-contraction coupling throughout the cell. Due to their structural complexity and narrow lumen, T-tub enclose extracellular fluid, allowing for only restricted diffusional exchange with bulk extracellular contents. To explore whether T-tub deformation during the cardiac contraction-relaxation cycle may alter T-tub shape and volume-surface ratio, we conducted dual-axis 3D ET (voxel size 1.2nm) of ventricular tissue fragments of rabbit hearts (n=6), fixed in different mechanical states (contracture, cardioplegic arrest, volume load). We then determined T-tub cross-section, eccentricity (deviation from circular cross-section), and the angle between maximal T-tub radius (rmax) direction and Z-disc plane and analysed it as a function of sarcomere length. We found that, as SL increases, rmax orientation changes from near-parallel to Z-disks in contractured cells to near-perpendicular during maximal stretch (P<0.001). In parallel, a bi-phasic change in T-tub eccentricity was seen, with higher values at short and long SL, and lower levels in-between (P<0.001). Surprisingly, the extend of T-tub volume-surface ratio changes as a result of mechanical deformation was less than expected, possibly due to dynamic in-/excorporation of caveolae as a function of T-tub membrane tension. Our findings show that T-tub undergo SL-dependent deformation, potentially associated with T-tub content mixing. Functional studies will investigate whether the changes in partial T-tub volumes could aid T-tub content exchange with the bulk extracellular space. We show that 3D ET yields accurate structural information that provides a structural framework for mechanistic integration of molecular signalling networks.