Key points Using 3D direct stochastic optical reconstruction microscopy (dSTORM), we developed novel approaches to quantitatively describe the nanoscale, 3D organization of ryanodine receptors (RyRs) in cardiomyocytes.Complex arrangements of RyR clusters were observed in 3D space, both at the cell surface and within the cell interior, with allocation to dyadic and non‐dyadic pools.3D imaging importantly allowed discernment of clusters overlapping in the z‐axis, for which detection was obscured by conventional 2D imaging techniques. Thus, RyR clusters were found to be significantly smaller than previous 2D estimates.Ca2+ release units (CRUs), i.e. functional groupings of neighbouring RyR clusters, were similarly observed to be smaller than earlier reports. Internal CRUs contained more RyRs in more clusters than CRUs on the cell surface, and yielded longer duration Ca2+ sparks. Cardiomyocyte contraction is dependent on Ca2+ release from ryanodine receptors (RyRs). However, the precise localization of RyRs remains unknown, due to shortcomings of imaging techniques which are diffraction limited or restricted to 2D. We aimed to determine the 3D nanoscale organization of RyRs in rat cardiomyocytes by employing direct stochastic optical reconstruction microscopy (dSTORM) with phase ramp technology. Initial observations at the cell surface showed an undulating organization of RyR clusters, resulting in their frequent overlap in the z‐axis and obscured detection by 2D techniques. Non‐overlapping clusters were imaged to create a calibration curve for estimating RyR number based on recorded fluorescence blinks. Employing this method at the cell surface and interior revealed smaller RyR clusters than 2D estimates, as erroneous merging of axially aligned RyRs was circumvented. Functional groupings of RyR clusters (Ca2+ release units, CRUs), contained an average of 18 and 23 RyRs at the surface and interior, respectively, although half of all CRUs contained only a single ‘rogue’ RyR. Internal CRUs were more tightly packed along z‐lines than surface CRUs, contained larger and more numerous RyR clusters, and constituted ∼75% of the roughly 1 million RyRs present in an average cardiomyocyte. This complex internal 3D geometry was underscored by correlative imaging of RyRs and t‐tubules, which enabled quantification of dyadic and non‐dyadic RyR populations. Mirroring differences in CRU size and complexity, Ca2+ sparks originating from internal CRUs were of longer duration than those at the surface. These data provide novel, nanoscale insight into RyR organization and function across cardiomyocytes.