Human breast milk is primarily colostrum immediately following birth. Colostrum gradually changes to mature milk over the next several days. The role of colostrum in fighting infections and promoting growth and development of the newborn is widely acknowledged. This role is mediated by differences across cultures in the acceptability of colostrum and the prevalence of colostrum feeding. This study examined the prevalence of colostrum feeding and time to initiation of breast-feeding in 143 rural Bangladeshi women in Matlab thana. Structured interviews were collected during a 9-month prospective study conducted in 1993. Women were usually interviewed within 4 days of giving birth and were asked about whether or not they fed their child colostrum and the number of hours until they began breast-feeding the baby. Ninety per cent of the mothers reported feeding their newborn colostrum. A logistic regression found no effect on the prevalence of colostrum feeding from the following covariates: mother's age, parity, history of pregnancy loss, child's sex, mother's self-report of delivery complications, and the time from birth to interview. Fifty-nine per cent of mothers initiated breast-feeding within 4 h, and 88% within 12 h of parturition. Survival analysis was used to estimate the effects of covariates on the time from delivery to initial breast-feeding. Time to initial breast-feeding was delayed slightly, but significantly, for older mothers, for male infants, and by mothers who did not report delivery complications. The percentage of mothers who fed their child colostrum was higher, and times to initial breast-feeding were shorter, than almost all previous reports from South Asia. These findings might be explained, in part, by methodological differences among studies, but it is suggested that recent changes towards earlier initiation of breast-feeding have taken place in rural Bangladesh.