Exercise electrocardiography is a valuable clinical tool with which quantitative evaluations and comparisons are now being attempted in many fields of cardiovascular investigation. Among a series of studies by a technical group on exercise electrocardiography, an assessment was made of observer variation in the clinical interpretation of ST-T responses during and after exercise. Interobserver variation among 14 cardiologists was great; individually assigned frequencies of abnormal responses after exercise in a mixed sample of records ranged from 5 to 58 per cent. Disagreement was greater for the diagnosis made during than for that made after exercise. Introbserver variation was also so great that use of the same observer for all exercise electrocardiographic readings would not necessarily provide acceptable reliability. The chief factors in disagreement were the lack of defined criteria for interpretation, in particular uncertainty about the significance of J-point ST-T depression, and technical quality of the records. Observer agreement was substantially increased when records were coded by unambiguous criteria or when simple measurements were made of the ST-T response after exercise.