U-series chronologies of the emerged coral limestone terraces on Barbados, West Indies, together with those of the terraces from New Guinea, have formed the basis for most late Pleistocene eustatic models. The so-called “Barbados sea level model” has been challenged in recent years, however. A major issue is whether during oxygen isotope stage 5e, when Rendezvous Hill reef complex on Barbados Island formed, the sea rose above the present position for one relatively brief period of <10,000 yr, or for two or more periods spanning approximately from 140,000 to 115,000 yr B.P. Evidence for the latter scenario has not come from initial studies of Barbados but from elsewhere; it is also inconclusive because of the dating uncertainties involved. We have carried out careful redeterminations of U-series ages on a suite of 29 Acropora palmata samples systematically collected from four of the lowest terraces on the island. Diagenetic disturbance may have caused the age spreads at some sampling outcrops. A model for the diagenetic exchange of uranium isotopes in coral samples with those in groundwater explains the anomalous 234U/238U ratios in samples with apparently unaltered mineralogy (aragonite) and trace element (Mg and Sr) chemistry. It shows that age dispersions of 5–10% can be engendered by a U exchange coefficient of the order of 10−6 yr−1. The lower-limit terrace ages, estimated from averaging the multiple measurements, are 81,000 ± 2000 yr (Worthing), 105,000 ± 1000 yr (Ventnor), 120,000 ± 2000 yr (Maxwell), and 117,000 ± 3000 yr (Rendezvous Hill). No evidence was found of previously inferred bipartite sea levels centering around 118,000 and 135,000 yr ago. This study documents the need of dating coral with the high precision/sensitivity mass-spectrometric techniques for future resolution of the temporal relationships among sea level changes, climate oscillations, and astronomical forcing—relationships originally addressed by the Barbados sea level model.