The purpose of this study was t o assess the practice of a group of non-nurse midwives by comparing the outcomes of their deliveries t o the stan- dard of hospital-based physicians. This was done by means of a matching procedure designed to equate the populations on major medical risk factors. The first phase of the study used aU physicians and found major differences favoring the midwives. The second phase of the study used the least interven- tionist half of the physician group and found minimal differences between midwives at home and physicians in the hospital. It is suggested that excessive intervention is not without risk regardless of who does it.