yd outside agencies which they do not have much control over (as has been the case with the Trust Territory Health Services Administration in the past). Thus, it is recommended that the proposed planning unit be administratively tied directly to the Ministry of Health Services and that it work closely with the Nitijela, the Health Coordinating Council (it if remains functional), and the Secretary of Health Services (see Organization Chart, Section IV A). As noted in the Organization Chart, the Planning/Evaluation Unit will be on the same level as the other main divisions. This will assume it has both the independence and clout needed to fulfill its role. It should be stressed that the planners (if expatriate) must be willing to adapt their planning methodologies to the traditional political system found in the M.I. Complicated Health Services Administration-type planning methodologies as used in the U.S. are not approriate here. health planning for developing The planners must have experience in countries and the support and backup by consultants of various categories will be needed, i.e. social scientists, survey statisticians and epidemiologists. 2. Manpower There is not available locally personnel who have the statistical, planning, and epidemiological expertise needed to adequately supervise the system. It is recommended that for at least five years a full-time expatriate planner/evaluator and a statistician/epidemiologist be hired. As soon as possible, promising Marshallese should be sent for advanced training in statistics and planning so that the expatriates can be phased out. There is also need for an expatriate hospital medical records technician to assist in revamping that system. This person could also assist in the planning/evaluation unit particularly in the design and upkeep of the clinicbased patient record keeping system. 3. Vital Statistics Gathering It is recommended that this unit make as its first priority the testing of other alternatives to the existing vital statistics and medical record system i.e. a patient-retained record system. The latter would be to counteract the double problems of a highly mobile population and the custom of frequent name changes. The possible utilization of the radio for selective data gatering should also be tested. In addition the design of a planned schedule of periodic sample surveys to assess program effectiveness needs to be done.