The lateral continuity and facies heterogeneities of metre-scale cycles in a greenhouse Lower Jurassic (Sinemurian) carbonate ramp from the northern Iberian Basin (Spain) was evaluated from extensive field analysis carried out on a well-exposed 12 km long outcrop. Eleven high-frequency continuous cycles and their bounding surfaces are traceable laterally through the entire outcrop. However, three of these cycles are found to split laterally into discontinuous cycles of more limited distribution (up to 3 to 5 km of lateral extent). The continuous and discontinuous cycles have a similar field expression in one-dimensional logs. As a consequence, the number of cycles that can be differentiated is variable along the logged sections (i.e. from 11 to 16). Cycles have variable facies heterogeneities and sedimentary trends depending on the environment of formation: shallowing-upward and symmetrical cycles occur in protected lagoon–tidal flat areas and in the open-marine, high-energy domain. These cycles show significant facies heterogeneities, which were controlled mainly by lateral migration of a mosaic of facies over an irregular topography. Deepening-upward and aggradational cycles are generated in low-energy, sub wave-base, open-marine areas. Facies are laterally homogeneous, reflecting low potential for carbonate accumulation and inability to fill the created accommodation space in this low-relief and relatively deep area. Cycle boundaries are generated by stages of rapid accommodation gain, involving the flooding of the carbonate ramp; they are more likely to originate from regional tectonic pulses (related to the extensional tectonics operating in the northern Iberian Basin) rather than greenhouse low-amplitude eustacy. Discontinuous cycles tend to occur in thickened areas and are interpreted as originating from the infill of wedge-shaped accommodation space resulting from differential subsidence (i.e. local tectonic pulses). In conclusion, where thickness variations occur in extensional settings lateral continuity of cycles should not be expected. In less well-exposed, or in one-dimensional sections and in wells, it would not be possible to distinguish continuous from discontinuous cycles, or to understand such two-dimensional heterogeneities. Identification of unique cycle-forming mechanisms or attempting cyclostratigraphic long-distance correlation of cycles is unrealistic without a detailed analysis of the architecture of cycles in laterally continuous outcrops.