Evidence is presented to indicate that placebos are far more effective in producing carefully defined relief of pathological pain than they are in the case of experimental pain. This is construed as further support for the view that placebos are more effective when stress is great than they are when stress is not so great. A similar situation holds for morphine. Certain drugs are effective in relieving visceral sensations only if an essential psychological state is present. This is, in effect, a new principle of drug action.