The frequency of oscillations in global ice volume as determined by dating of coral reef deposits provides an important test of the Milankovitch model of orbital climate forcing. The timing of two episodes of high sea level during marine isotope substages (MIS) 5a and 5c, here determined to have occurred at ∼84 and ∼101 ka BP are broadly consistent with the Milankovitch model. However, new high precision U-series dating of uplifted morphologically distinct coral-reef terraces at Barbados indicate that MIS-5a was both longer and more complex than previously established, with an additional sea-level highstand occurring at ∼77 ka BP. The U-series analysis of a large number of samples for each coral reef deposit reveals a well-defined correlation between measured age and initial δ234U. This correlation is evident even within the uncertainty band of ±8‰ commonly used to exclude diagenetically altered samples and allows a more precise and accurate constraint of the age of these deposits by evaluating where the trend line intersects the modern δ234U value. Other proxy climate records also indicate suborbital-period variability during MIS-5a and 5c and together imply that, as for the last ice age, orbital forcing alone cannot account for the observed changes in climate.