Objective: This comparative study evaluated the readability and test-retest reliability of a questionnaire designed to assess the attitudes, beliefs behaviours and sources of information about HIV/AIDS among young adults recruited from universities in the United States of America (USA), Turkey and South Africa. Design/Setting: The instrument was administered on two occasions, within a two week interval, to 219 university students in the USA (n = 66), Turkey (n = 53) and South Africa (n = 100). Method: The psychometric instrument developed has five major subscales: demographic, HIV/AIDS attitudes and beliefs, HIV risk sexual behaviour, alcohol and drug use, and HIV sources of information. Results: The instrument's readability evaluation revealed a Flesch-Kincaid score (literacy difficulty level of the questionnaire) of 8.4, indicating that respondents would need an eighth grade reading level to understand the survey. The overall test-retest reliability coefficients for the items on the demographic subscale were generally high (0.893-0.997). Similarly, high test-retest reliability was obtained for the HIV risk sexual behaviour (0.738-0.996) and the alcohol and drug use (0.562-1.000) subscales. Much lower test-retest reliability was obtained for the HIV/AIDS attitudes and beliefs (0.32-0.80), and sources of information about HIV/AIDS (0.370-0.892) subscales. Conclusion: We found no discernible difference in the reliability data among the respondents from the three countries. The instrument should be of interest to clinicians and researchers investigating the HIV risk behaviours of young adults and older age groups with an eighth grade reading level. The availability of this instrument may enhance HIV population and intervention studies internationally. (Contains 5 tables.)