Stigma reduction programs are dominated by a biomedical model that presents depression as a medical illness. Alternately, a contextual model emphasizes that one should not be blamed for environmental influences. This study compared biomedical, contextual, and control stigma reduction programs to each other and to a no-program control. The main hypotheses were that the contextual program would have the greatest impact and that a match between participants' beliefs about depression and the model presented would moderate this effect. Seventy-four participants were randomized to the 3 programs and 12 participants served as a no-program control. The contextual and control programs reduced stigma significantly compared with the no-program control, whereas the biomedical program did not. Beliefs about depression moderated this effect only for the biomedical condition. Contextual and control programs seem to be effective but a biomedical model may be risky for those who disagree with the model. Theoretical implications are discussed.