The 'malaria hypothesis' predicts that the frequency of the sickle allele, which is high in malaria-endemic African populations, should decline with each generation in populations of African descent living in areas where malaria is no longer endemic. In order to determine whether this has been the case in Jamaica, we compared haemoglobin electrophoresis results from two hospital-based screening programmes separated by more than 20 years (i.e. approximately one generation). The first comprised 100,000 neonates screened between 1973 and 1981, the second, 104,183 neonates screened between 1995 and 2003. The difference in frequency of the sickle allele was small (5.47% in the first cohort and 5.38% in the second screening cohort) and not significant (Z = 1.23, P = 0.22). The same was true of the sickle trait frequency (10.05% in the first cohort and 9.85% in the second, Z = 1.45, P = 0.15). These differences were smaller than predicted under simple deterministic models based on the malaria hypothesis, and suggest that these models may not capture important determinants of allele and trait frequency decline (or persistence) in contemporary populations. Refining the expectations for allele and trait frequency change for Jamaica and other similar populations is an area for future study.