In the United States, blood donors have been routinely screened for human T-cell lymphotropic virus (HTLV) since 1988. HTLV-I and -II seropositive blood donors have been identified through confirmatory testing at five participating blood centers and frequency-matched seronegative controls provided information on potential HTLV sociodemographic, parenteral, and sexual risk factors during structured interviews. After adjustment, low educational attainment; accidental needlesticks or cuts; prior blood transfusion; > or = 7 sex partners; and a sex partner from an HTLV-I endemic area were significantly associated with both HTLV-I and -II. Gender did not modify the odds ratios (OR) in the final logistic regression models, despite apparent male-female differences in gender-specific bivariable analysis. Injection drug use (IDU) or having sex with an IDUs were significant risks for HTLV-II, but not for HTLV-I. The OR for donors who had IDU sex partners was 20.6 times higher than those who did not. For IDUs, the OR was increased 10.5 times over nonusers. Abortion was a significant HTLV-II risk factor for women. Our findings indicate that IDU and sex with IDUs are important risk factors for HTLV-II transmission, even among low-risk populations such as blood donors.