In outcrops, shallow-water carbonates often form thick, vertically stacked, metre-scale, exposure-bounded depositional packages, here called carbonate cycles. Because the facies within carbonate cycles can be tied precisely to water depths, they are considered ideal for reconstructing past sea-level changes. There is, however, increasing evidence that these depositional cycles are unreliable recorders of both the frequency and the amplitude of orbitally driven sea-level fluctuations. Carbonate cycles record only a fraction of the amplitude of a sea-level cycle: a portion of the rise and nothing of the fall. Cores through the Pleistocene cycles on Great Bahama Bank illustrate this shortcoming. Although the amplitudes of the last nine sea-level changes are each a hundred metres or more, the thickness of the cycles varies from a few metres to ca 15 m. The lack of correlation between sea-level amplitude and cycle thickness is not eliminated for cycles deposited during times of lower sea-level amplitudes, for example, the Cretaceous. Upper Cretaceous cycles on the Maiella platform margin document the irregularly filled accommodation space and the resultant variability in cycle thickness and frequency. Uncertainties in assessing the frequencies of sea-level changes from shallow-water carbonate cycles are caused by ‘missed beats’ and metre-scale oscillations of sea-level within highstands that potentially produce cycles of very short duration. The random amplitude variability during the last 57 glacio-eustatic sea-level changes illustrates the difficulty of assessing ‘missed beats’, where a sea-level fluctuation is not recorded because the sea-level rise does not reach the platform top. ‘Missed beats’ are also produced by the depositional topography that is created by irregularly filled accommodation space. As a result, variable numbers of cycles are deposited across the platform. Further complicating orbital frequency analyses are decametre-scale oscillations of the sea-level during highstands. The amplitude of these sub-orbital sea-level oscillations (up to 17 m within the last interglacial, Marine Isotope Stage 5e) are sufficient to expose shallow platforms like Great Bahama Bank and subsequently produce an additional depositional cycle with similar facies successions. The combined effects of missed beats and oscillations within highstands are likely to produce cycles and hiatuses of variable duration that are difficult to extract from the rock record. Consequently, estimates of the orbital forcing mechanisms and frequencies from ancient shallow-water carbonate cycles carry large uncertainties.