Considerable numbers of measles cases occur below the target age for vaccination in the Indian sub-continent. The immunogenicity of measles vaccine in infancy is dependent on the rate of decay in maternal antibody since this antibody interferes with vaccine induced seroconversion. This study investigated maternal antibody decay in a rural population in Bangladesh and evaluated possible risk factors for early decay. Measles antibodies were assessed using both ELISA and Plaque Reduction Neutralization (PRN) test in 330 infant-mother pairs in a cross-sectional survey. PRN was more sensitive method than ELISA for determining antibody levels. Antibody levels decreased rapidly in infants with increasing age. By the age of 5 months, 67% (28/42) infants had practically no protective antibody left (30 mIU ml-1 or below). Only 12% infants at 5 months of age, and 5% at 8 months, had levels greater than 120 mIU ml-1--stated to 'protect' children. Multiple regression showed that maternal age was the only variable associated with the level of antibody (maternal weight, height and MUAC were not associated), decreasing by 1.06 mIU ml-1 for each year of age (P = 0.002). Infant's antibody concentration decreased with age by an average 2 mIU mL-1 for every month of life (P < 0.0001), and was determined by the maternal antibody concentration (P < 0.0001) (child's length, weight, MUAC, mother's gestational age and parity were not associated). The relatively rapid antibody decay suggests that the target age for measles vaccination might be reduced. Further, as the cohort of vaccinated mothers enters reproductive age in Bangladesh, a more rapid decay of antibody may be expected in future generations of Bangladeshi children. The information presented here suggests that a formal trial of standard measles vaccine at younger ages is justified in this population as it could confer considerable benefit in reducing infant measles.