The impact of an on-line computer scheduling of patient care activities (implemented at the Texas Institute for Rehabilitation and Research) was measured from the standpoint of: (1) effectiveness (personnel awareness of rehabilitation goals for the patient, opinion regarding 18 factors of patient management, and patient opinion of the scheduling); (2) efficiency (schedule of activities actually carried out, speed in processing orders, absenteeism from specialized therapy); and (3) efficiency of computer operations (computer time, cost, and terminal usage). From the standpoint of effectiveness, the study indicated that the computerized system increased goal agreement between nursing personnel. It was rated very favourably by representatives of all disciplines involved in rehabilitation and by the patients themselves. From the standpoint of efficiency, there was an increase in the number of scheduled activities actually carried out, although not as much as it had been predicted. The time required to implement initial orders after admission decreased from approximately 2 hr to several minutes, but there was still a time lag between the time an order was written to the time it was communicated to the departments not involved in nursing care. The system did not have any effect on absenteeism from physical or occupational therapy. From the standpoint of efficiency of computer operations, the study shows that a plateau has been reached in teleprocessing time, programmer man hours, data-entry man hours, and batch-processing time. This computer system offers great potential in providing an organized, coordinated, and concise profile of the medical, physiological, psychological, and socioeconomic needs of the patient. The study points out the complexity of evaluative studies of the effect of computer systems in hospitals. In spite of the difficulties, such studies are necessary if we are to increase delivery of services within the constraints of limited available resources.