Thirty-one new bulk-sediment U-Th dates are presented, together with an improved δO stratigraphy, for ODP Site 1008A on the slopes of the Bahamas Banks. These ages supplement and extend those from previous studies and provide constraints on the timing of sea-level highstands associated with marine isotope stages (MIS) 7 and 9. Ages are screened for reliability based on their initial U and Th isotope ratios, and on the aragonite fraction of the sediment. Twelve 'reliable' dates for MIS 7 suggest that its start is concordant with that predicted if climate is forced by northern-hemisphere summer insolation following the theory of Milankovitch. But U-Th and δO data indicate the presence of an additional highstand which post-dates the expected end of MIS 7 by up to 10 ka. This event is also seen in coral reconstructions of sea-level. It suggests that sea-level is not responding in any simple way to northern-hemisphere summer insolation, and that tuned chronologies which make such an assumption are in error by ≈10 ka at this time. U-Th dates for MIS 9 also suggest a potential mismatch between the actual timing of sea-level and that predicted by simple mid-latitude northern-hemisphere forcing. Four dates are earlier than that predicted for the start of MIS 9. Although the most extreme of these dates may not be reliable (based on the low-aragonite content of the sediment) the other three appear robust and suggest that full MIS 9 interglacial conditions were established at 343 ka. This is ≈8 ka prior to the date expected if this warm period were driven by northern-hemisphere summer insolation.