In order to shed light on the mode of HTLV-I infection by mother-to-child transmission, we examined sera of school children in a highly endemic town on two separate occasions at a 6-year interval. The carrier rates in ages 15-17, 8.7 and 2.1%, were significantly higher than that in ages 6-8, 1.7 and 0.4%, in studies. The latter survey showed a significantly lower carrier rate in each age group. Moreover, the carrier rates of those students born in 1965-1967 and 1968-1970 were stable in the interval. The data suggested that carrier rates of children at certain ages are reflected by the date of birth rather than by age. A prospective survey of children born of carrier mothers found the overall carrier rate to be approximately 25%, which did not increase with their age. There was no sexual difference in the carrier rate of children: 5/25 in male and 9/34 in females (X2 = 0.3). Carrier mothers could be separated into two groups: HTLV-I antigen-positive mothers and negative mothers. Nine out of 19 antigen-positive mothers (47%) and 2 of 19 antigen-negative mothers (11%) had carrier children (X2 = 6.3). Twelve of 30 children born of antigen-positive carrier mothers (40%) were carriers, in contrast to 2 of 24 children (8%) of antigen-negative mothers (X2 = 7.8). Furthermore, 12 of 14 carrier children (86%) were of antigen-positive mothers. This suggests that postnatal but early transmission of HTLV-I plays a significant role in the maintenance of HTLV-I endemy and the development of adult T-cell leukemia/lymphoma.