OBJECTIVE: To reconstruct the historical changes in force of dengue infection in Singapore, and to better understand the relationship between control of Aedes mosquitoes and incidence of classic dengue fever. METHODS: Seroprevalence data were abstracted from surveys performed in Singapore from 1982 to 2002. These data were used to develop two mathematical models of age seroprevalence. In the first model, force of infection was allowed to vary independently each year, while in the second it was described by a polynomial function. Model-predicted temporal trends were analysed using linear regression. Time series techniques were employed to investigate periodicity in predicted forces of infection, dengue fever incidence and mosquito breeding. FINDINGS: Force of infection estimates showed a significant downward trend from 1966, when vector control was instigated. Force of infection estimates from both models reproduced significant increases in the percentage and average age of the population susceptible to dengue infections. Importantly, the year-on-year model independently predicted a five to six year periodicity that was also displayed by clinical incidence but absent from the Aedes household index. CONCLUSION: We propose that the rise in disease incidence was due in part to a vector-control-driven reduction in herd immunity in older age groups that are more susceptible to developing clinical dengue.