OBJECTIVE The authors examine 58,441 obstetric deliveries in New York State outside New York City to test for the existence of defensive medicine in obstetrics. METHODS The data consist of merged vital statistics and hospital discharge records from the New York State Department of Health, together with other merged variables. Physician fear of malpractice is proxied by cumulative obstetric malpractice suits by county for 1975 through 1986. A generalized probit analysis is used. RESULTS Malpractice exposure is shown to influence slightly the use of the electronic fetal monitor (EFM), a major diagnostic tool. Use of the EFM is shown to influence the diagnosis of fetal distress; fear of malpractice influences this diagnosis both directly and through the EFM. The diagnosis of fetal distress significantly affects the choice of cesarean section (c-section) as a method of delivery; hence, fear of malpractice influences the choice of a c-section both directly and through the diagnosis of fetal distress. Failure to include indirect effects via diagnostic procedures and diagnosis would result in an underestimate of the effect of fear of malpractice. Of an overall c-section rate of 27.6% in the data set, fear of malpractice accounts for an estimated 6.6 percentage points, of which 4.4 percentage points reflect a direct effect, and the remaining 2.2 percentage points reflect the effect of malpractice exposure on the use of the EFM and, directly and indirectly, the diagnosis of fetal distress. CONCLUSIONS The results appear to confirm the existence of defensive medicine in obstetrics. Whether this is a desirable or undesirable effect remains ambiguous, but it is costly.