This paper estimates the effect sizes of placebo treatment of depression from meta-analyses of antidepressive medicine, and examines methodological artifacts in antidepressive research. It is concluded that between one-half and two-thirds of the effects of antidepressive medicine are explained by placebo response with a mean difference between treatment and placebo groups around 25 percent. This modest effect size should probably be adjusted for methodological bias due to incomplete blindness and other sources of error favouring the medicine group. Some practical consequences of the fact that depression is a placebo sensitive disorder are outlined.