Assessing the impact of new technologies on health care costs is an important research area. This paper evaluates two technologies used to treat coronary artery disease. We estimate two separate stochastic frontier models—one for balloon angioplasty patients and one for cardiac bypass surgery patient—using data taken from detailed chart and cost files of a large urban hospital. Cost estimates for the two technologies are purged of inefficiency so that forecasts of the cost consequences of technology shift can be based on ‘best-practice’ production techniques. Learning behavior, physician effects, and patient clinical characteristics are also taken into account. We find that there are potential cost savings associated with making angioplasty a more perfect substitute for bypass surgery, as well as current inefficiency in production.