Evidence of the cost-benefits and cost-effectiveness of medical rehabilitation is reviewed. In stroke rehabilitation, favorable evidence was found, but also the contrary. For spinal cord injury, studies of effectiveness and costs have shown the value of specialized rehabilitation centers. For other areas of medical rehabilitation, investigations have given hints regarding cost-effectiveness but little firm evidence. In allied areas--rehabilitative alternatives in long-term care, vocational rehabilitation, and medical care in general--studies were identified that hold valuable lessons for future study of cost-benefits. A concluding critique identifies obstacles to the study of cost-benefits and makes specific suggestions. Improved experimental designs are required to control for selection biases. Alternatives to ongoing practice need to be entertained if cost-benefit study is to be useful.