Hodder R A (Department of Preventive Medicine and Biometrics, Uniformed Services University of Health Sciences, 4301 Jones Bridge Road, Bethesda, Maryland 20814, USA), Longfield J N, Cruess D F and Horton J A. Teaching concepts of clinical measurement variation to medical students. International Journal of Epidemiology 1982, 11: 287-292., An exercise in clinical epidemiology was developed for medical students to demonstrate the process and limitations of scientific measurement using models that simulate common clinical experiences. All scales of measurement (nominal, ordinal and interval) were used to illustrate concepts of intra-and interobserver variation, systematic error, recording error, and procedural error. In a laboratory, students a) determined blood pressures on six videotaped subjects, b) graded sugar content of unknown solutions from O to 4 using Clinitest tablets, c) measured papules that simulated PPD reactions, d) measured heart and kidney size on X-rays and, e) described a model skin lesion (melanoma). Traditionally, measurement variation is taught in biostatistics or epidemiology courses using previously collected data. Use of these models enables students to produce their own data using measurements commonly employed by the clinician. The exercise provided material for a meaningful discussion of the implications of measurement error in clinical decision-making.