ESR is a time-honored, simple, inexpensive test, but unfortunately it lacks sensitivity and specificity. Clinicians need to be aware of appropriate uses, because any test is expensive when ordered often, and evaluation of false-positive results may incur substantial costs and place the patient at risk from additional procedures. ESR should not be used to screen asymptomatic persons for disease. If an increased ESR is encountered and no explanation is immediately apparent, clinicians should repeat the test in several months rather than pursue an exhaustive search for occult disease. ESR may be useful in establishing a "sickness index" in elderly persons who have nonspecific changes in health status and a moderate probability of underlying disease; in screening for infection in specific settings (e.g., orthopedic surgery, pediatrics, gynecology); in diagnosing and monitoring temporal arteritis, polymyalgia rheumatica, and possibly other rheumatic diseases; in monitoring patients with treated Hodgkin's disease; and in assessing iron deficiency in anemia of chronic disease (when correlated with serum ferritin level). An ESR value exceeding 100 mm/hr has a 90% predictive value for serious underlying disease, most often infection, collagen vascular disease, or metastatic tumor. In asymptomatic persons with a markedly elevated ESR value, a minimal number of tests usually reveal the cause.