Briefly trained physicians assistants using protocols (clinical algorithms) for diabetes, hypertension, and related chronic arteriosclerotic and hypertensive heart disease abstrated information from the medical record and obtained history and physical examination data on every patient-visit to a city hospital chronic disease clinic over a 18-month period. The care rendered by the protocol system was compared with care rendered by a "traditional" system in the same clinic in which physicians delegated few clinical tasks. Increased thoroughness in collecting clinical data in the protocol system led to an increase in the recognition of new pathology. Outcome criteria reflected equivalent quality of care in both groups. Efficiency time-motion studies demonstrated a 20 per cent saving in physician time with the protocol system. Coct estimates, based on the time spent with patients by various providers and on the laboratory-test-ordering patterns, demonstrated equivalent costs of the two systems, given optimal staffing patterns. Laboratory tests were a major element of the cost of patient care,and the clinical yield per unit cost of different tests varied widely.