In order to evaluate the protective role of the maternal antibody against mother-to-child transmission of HTLV-I, we followed a total of 780 children born to HTLV-I carrier mothers by investigating the level of anti-HTLV-I antibody transferred in utero, decline of the maternal antibody and seroconversion in post-natal life. The anti-HTLV-I antibody was positively detected within the first 3-6 months of life and declined at 6-12 months after birth in all children. After the maternal antibody declined, seroconversion occurred in some of the children following either breast feeding or bottle feeding. The seroconversion rates of short-term (less than or equal to 6 months) and long-term (greater than or equal to 7 months) breast feeders were 4.4% (4/90 cases) and 14.4% (20/139 cases), and the rate of bottle feeders was 5.7% (9/158 cases). Long-term breast feeding yielded more seroconverters than short-term breast feeding; 14.4% (20/139 cases) vs. 4.4% (4/90 cases), RR = 3.68, p = 0.018. The seroconversion rate of short-term breast feeders was nearly equal to that of bottle feeders; 4.4% (4/90 cases) vs. 5.7% (9/158 cases), RR = 0.770, p = 0.471. When neonatal lymphocytes were cultured with breast milk cells of HTLV-I carrier mothers, the in vitro infection of HTLV-I was inhibited by the addition of HTLV-I-seropositive cord-blood plasma. Our results suggest that the maternal antibody may inhibit HTLV-I infection by short-term breast feeding but not by long-term breast feeding after decline of the maternal antibody.