A fundamental question in the correlation of 1-D sedimentologic data is whether to use a layer-cake or shingled correlation approach. The resulting reservoir geometry has important implications for the characterization of reservoir heterogeneities and fl uid fl ow. On the Saiq Plateau in Oman, epeiric carbonate ramp deposits of the Triassic Sudair Formation are well exposed and can be investigated in detail over several kilometers. There, reservoir heterogeneities on different scales have been documented by creating various outcrop wall panels and 2-D correlations. Multi-level architectural elements with different depositional geometries were discovered, which were linked to a sequence-stratigraphic hierarchy consisting of three levels. Level 1: A "layercake"- type stratigraphic architecture with minor thickness variations over several kilometers becomes apparent when correlating fourth-order cycle set boundaries. Level 2: The correlation of fifth-order cycle boundaries refl ects horizontally continuous geometries, within which, however, internal grainstone layers were discovered to be arranged in a shingled fashion. Muddy layers in between these shingles illustrate sixth-order mini-cycle boundaries. Level 3: Within sixth-order mini-cycles another scale of a shingle-like architecture can be observed. Amalgamated cm-thick grainstone units form thin wedges with subtle but clearly inclined dipping geometry. Fourth-order cycle sets and fifth-order cycles can be traced over several kilometers, and therefore assumed to be related to allocyclic stratigraphic processes. The internal shingle geometries within fifth-order cycles are traceable over 100s of meters and presumably refl ect an autocyclic lateral migration of a shoal complex. Cm-thick shingling grainstone wedges within sixth-order mini-cycles are interpreted as storm-related spill deposits. Their event-driven character is refl ected by frequent amalgamation and reworking of the preceding deposits. The results of this study of epeiric carbonate ramp deposits suggest that a "layer-cake" correlation approach is appropriate when correlating 10s of m-thick grainstone units over a distance of several kilometers. However in the documented example, these thick grainstone units consist internally of smallscale architectural elements, which show inclined geometries and require a shingled correlation approach. These small-scale heterogeneities within an overall "layer-cake" architecture might have an impact on fl uid fl ow in similar subsurface reservoirs and should be taken into account for detailed reservoir correlations and static reservoir models.