Projection of the probable demand for health services over time is one of the most important–and difficult–aspects of the health planning process. Not only must the planner contend with many variables external to the health sector, but also with the difficult to predict correlations between the supply of health services and the resultant demand. This article briefly reviews some of the principal demand projection methods in use and then describes in detail the one adopted in the Chilean health manpower study. The demand portion of the study in Chile involved three main elements: 1) a sample survey which measured the met and unmet demand for medical, dental, and hospital services as a function of six population variables (age, sex, location, income, educational level, and medical insurance status); 2) a baseline demand projection which takes into account the probable effects of changes in these six variables over ten and 20 years on the utilization of services; and 3) an alternative projection which postulates the fulfillment of certain targets for the improvement of health care. The approach offers the planner a number of important analytical and programmatic advantages compared with other methods now available and are discussed along with their limitations.