Cardiac arrhythmias are often driven by defects in electrical impulse conduction and are common to multiple pathologies, including heart failure and cardiomyopathy. However, the structural substrate of arrhythmias has proven difficult to study or influence in an experimental setting, and even more so in the clinic. Emerging evidence suggests that structural substrates for arrhythmias can exist at subcellular spatial scales extending down to the nanometre level. Such evidence derives mainly from experimental and modelling studies of the intercalated disk (ID), sites of cell–cell contact. The ID is home to gap junctions (GJ) and cardiac voltage-gated Na+ channels (NaV1.5), which are considered to be critical determinants of electrical conduction in cardiac tissue. The GJ and Na+ channel proteins are known to cluster and form nanodomains within regions of narrow intermembrane separation called the perinexi. Disruption of perinexi and the ID is associated with increased risk for arrhythmias: experimental perturbation of ID nanodomains induces proarrhythmic conduction defects, and disruption of perinexal nanodomains has been identified in human arrhythmia patients (Raisch et al. 2018). However, limitations of current technology preclude direct functional investigation of cardiac impulse propagation at the nanoscale. Thus, for the foreseeable future, computational models will be uniquely capable of tackling such questions. While conventional cardiac tissue modelling approaches typically neglect ID structure and Na+ channel clustering, and therefore cannot represent these ID perturbations, earlier studies have modelled simplified representations of the ID. Kucera and colleagues identified a critical mechanism in which Na+ channels localized at the ID impact conduction: ID-localized Na+ currents hyperpolarized the intercellular cleft (the narrow extracellular space between coupled myocytes), which depolarized the post-junctional ID membrane and activated downstream Na+ current (Kucera et al. 2002). These interactions occurring in the extracellular space are collectively termed ephaptic coupling. This and other studies showed that ephaptic coupling can impact conduction in a complex manner, dependent on structural and tissue properties including GJ coupling, cleft width, and ID Na+ channel localization (Weinberg, 2017). Yet these early works still considered the ID as a homogeneous structure, neglecting the significant heterogeneity that arises due to ion channel clustering around GJs. Hichri and colleagues illustrated a pivotal advance: 2D finite-element modelling of the ID membrane and cleft space demonstrated that the size and location of Na+ channel clusters within the ID could also significantly impact ephaptic coupling (Hichri et al. 2018). In this issue of The Journal of Physiology, Ivanovic and Kucera further extend this work to consider important structural details, specifically the effects of Na+ channel clustering around the GJ plaque and location within the narrow perinexus region (Ivanovic & Kucera, 2021). The authors developed a 3D high-resolution finite-element model of two apposing cells and a GJ plaque-adjacent Na+ channel cluster, in which they rigorously investigated the relationship between intermembrane separation and the relative location of Na+ channels, the perinexus, and the GJ plaque, along with the resulting propagation delay between cells. The authors demonstrated that the location of the Na+ channel clusters both near GJs and within a narrow perinexus were critical factors in facilitating impulse transmission. Transmission between cells was successful for a wide range of conditions when Na+ channels were located near GJs within a narrow perinexus. In contrast, transmission failed for nearly all conditions when Na+ channels were outside of the narrow perinexus. These important structural details have important implications for pathological conditions, in which ID structural remodelling results in Na+ channels translocating away from GJ plaques and widening of the perinexi. This study required the development of sophisticated numerical methods and model reduction techniques, which facilitated a more detailed investigation of mechanisms underlying ephaptic coupling that would not have been possible with a simpler model. The authors considered a novel hypothesized two-step interaction occurring in ephaptic coupling, specifically, that (1) inward Na+ current on the pre-junctional ID membrane can, upon cell entry, flow via GJs into the post-junctional cell, and (2) subsequently activate Na+ channels on the post-junctional ID membrane. Interestingly, simulations identified conditions in which the first step occurred, but in general the second step did not occur. Additionally, Ivanovic and Kucera identified a new mechanism involved in ephaptic coupling, termed ‘Na+ transfer’. During impulse transmission, Na+ current in the pre-junctional ID membrane is briefly outward, which acts as a current source for Na+ current flowing into the post-junctional cell. This study represents an important advance in understanding the functional role of the ID and the clinical significance of ID disruption and ion channel organization. The authors acknowledge limitations of their approach, in particular the simplified representation of the overall ID geometry. The authors assumed a flat ID, while noting the known highly tortuous morphology identified in microscopy studies, and also simulated a single Na+ channel cluster and GJ plaque. Our group recently developed a finite-element-based approach to study the effects of heterogeneous ID geometry (e.g. distinct plicate and interplicate regions) and region-specific GJ plaque sizes, and found that these properties also impacted conduction by modulating ephaptic coupling (Moise et al. 2021). Thus, there is a critical need to develop new approaches that can interrogate the role of structural detail across a wide range of spatial scales: computational models integrating high-resolution Na+ channel clusters, heterogeneous ID structure, and tissue-scale propagation; and experimental tools that can identify, quantify, and perturb ID structure and ion channel organization. These tools and techniques will ultimately provide a deeper understanding of the link between cardiac ID structure and conduction. This concept, once established, will be the gateway to new diagnostic tools which directly investigate ID structure and new therapies that can modulate or repair it and thus improve cardiac conduction. No competing interests declared. All authors have read and approved the final version of this manuscript and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed. This perspective was supported by funding from National Institutes of Health grant numbers R01HL138003 (to S.H.W.) and R01HL148736 (to R.V.).