This article reviews research from the five income-maintenance experiments in Canada and the United States. After sketching the historical and political context of the experiments, we compare their designs and discuss some important analytic difficulties. Our primary focus is the work-incentive issue, both nonstructural estimates of the experimental effects and elasticity estimates of structural labor-supply functions. We provide initial estimates of nonstructural and structural models for the Canadian experiment. We discuss more briefly some non-work-response findings associated with a guaranteed annual income and offer some personal comments on social experimentation and the policy process.