Audio computer assisted self-interview (ACASI) has been shown to elicit significantly higher reports of sensitive HIV risk behaviors compared with interviewer-administered questionnaires (IAQ). Injection drug users were randomized to either ACASI (n = 556) or IAQ (n = 586) in order to evaluate reporting differences. A significant interaction was found between mode of administration (ACASI/IAQ) and HIV status. HIV-seropositive respondents on ACASI compared with IAQ were less likely to report condom use (odds ratio [OR] = .42, p < .01) and condom use frequency (OR = .28, p < .002), compared with HIV-seronegative respondents. Based on ACASI results, HIV-seropositive individuals engaged in fewer preventive behaviors than suggested by IAQ. These findings support the validity of ACASI as an important mode of data collection in reducing socially desirable responding, especially among HIV-seropositive subjects.