IntroductionVertical transmission of human T-lymphotropic virus type I (HTLV-I) depends primarily on breast-feeding; substitution of bottle-feeding has reduced the transmission rate from 20% in breast-fed children to 3% among bottle-fed. To determine the correlates of transmission for long breast-feeding (2 6 mo), short breast-feeding (< 6 mo), and bottlefeeding mothers, the antibody titers of transmitter (T) mothers and non-transmitter (nT) mothers were analyzed by using synthetic and recombinant epitopes representing the immunodominant epitopes of gag (Gagla, r24), env (Envl/5, MTA1, RE3), and tax (Tax8/22-24) proteins.Seroreactivity to gag and tax epitopes was not significantly different except for anti-r24 antibody titer, which was significantly higher among T-mothers (geometric mean 134) when compared with nT-mothers (62) in the long-feeding group (P < 0.001). Profiles of antibody titers against env epitopes were different. Within the long-feeding group, Envl/5, MTA1, and RE3 titers were significantly higher among T-mothers (258, 1,476, and 738, respectively) when compared with nT-mothers (106,279, and 320, respectively) (P < 0.01 for all three epitopes). In contrast, within the bottle-feeding group, antibody titers to Envl/5 (269) and RE3 (418) among nT-mothers were significantly higher than those among T-mothers (80 and 113, respectively) (P < 0.01). These data confirm that high-titered anti-HTLV-I antibodies in the long-feeding group correlate with milkborne transmission of HTLV-I and, more importantly, imply that maternal anti-env antibodies may reduce the risk of non-milkborne infection. (J. Clin. Invest. 1995Invest. . 95:2920Invest. -2925