Where immunization campaigns locally eliminate measles, it will be imp
ortant to identify the vaccination policy most likely to prevent futur
e epidemics. The optimum age for vaccination depends on the rate of de
cline of maternal antibody, because the presence of antibody reduces v
accine efficacy. The first part of this paper contains a quantitative
reappraisal of the data on antibody decline and seroconversion rates b
y age. The decline in maternal antibody protection follows delayed exp
onentials, with delays of 2-4 months, and subsequent half-lives of 1-2
months. Using this result in an analytical mathematical model me find
that. the optimal age to administer a single dose of vaccine to child
ren, which is independent of vaccine coverage, lies within the range 1
1-19 months. We also show that, where the optimal age cannot be met, i
t is better to err towards late rather than early vaccination. There a
re therefore two reasons why developing countries, which presently vac
cinate during infancy because measles transmission rates are high shou
ld eventually switch to the second year of life. The possible gains fr
om two-dose vaccination schedules are explored with respect to both co
verage and efficacy. A two-dose schedule will be beneficial, in princi
ple. only when there is a need to increase net vaccine efficacy, after
coverage has been maximized with a one-dose schedule.