The high frequency of the sickle allele in some parts of Africa is understood to be a consequence of high malarial endemicity. One corollary of this is that the sickle allele frequency should be declining in populations of African ancestry that are no longer exposed to malaria. We have previously shown that there has been no change in sickle allele frequency in malaria-free Jamaica between two large-scale neonatal screening exercises conducted in 1973-1981 and 1995-2003. To evaluate the determinants of, and derive expected values for, sickle allele frequency in Jamaica, local empirical data were used to estimate the parameters of deterministic models of allele frequency decline. We found that although model predictions were broadly consistent with observed values in the 1973-1981 cohort, the predicted change in allele frequency between the two cohorts was larger than the observed, nonsignificant, reduction. Close agreement between predicted and observed values was only achieved by simulating a recent, marked increase in HbSS fitness. Thus, the "unexpected" persistence of the sickle allele in Jamaica may reflect the fact that the actual fitness among SS individuals is higher than that previously realized. If true, our models suggest that without substantial changes in current screening and counseling practice, there will be little "natural" reduction in sickle allele frequency for several hundred years. Better estimates of relative fitness will be helpful in refining these predictions and may aid in assigning health care priorities in Jamaica and the African Diaspora.