Background: One hundred and twenty educators from KwaZulu-Natal, South Africa, underwent HIV/AIDS training. The educators were surveyed about their attitudes toward people with HIV. Methods: The educators completed self-administered survey questionnaires both before and after 2 interventions. Measures included demographic characteristics, teachers' knowledge about HIV/AIDS, self-efficacy in handling HIV/AIDS situations, and attitudes (stigma and otherwise) toward HIV-related issues. The first intervention was a CD-ROM and the second intervention involved educators receiving a 2-day workshop on HIV transmission, risk factors, and actions that educators should know and undertake. The first step entailed testing the stigma instrument for its internal consistency and developing and testing potential subscales from the instrument. The second step entailed testing for the statistical associations between stigma (as measured by the stigma instrument and its subscales) and various demographic and HIV knowledge-related variables. Results: The overall stigma scale had a Cronbach [alpha] coefficient of 0.66. Educators in the workshop generally had lower baseline levels of stigma than those in the CD-ROM intervention. Following both interventions, the stigma levels of both groups of educators were significantly reduced. Improvements in the general knowledge about HIV as a disease and in the specific knowledge about HIV transmission risks were both statistically significantly associated with reductions in stigma attitudes, having controlled for educators' baseline stigma levels and demographic profiles. Conclusions: The levels of teachers' stigma attitudes were statistically significantly lower after both types of HIV/AIDS training and were also statistically significantly associated with improvements in HIV knowledge. (Contains 4 tables.)