To investigate the influence of physicians' practice styles on the rate of deliveries by cesarean section, we studied 1533 affluent women at low risk of obstetrical complications who were cared for by 11 obstetricians in a single community hospital. The mean rate of delivery by cesarean section was 26.9 percent, but the rate ranged from 19.1 to 42.3 percent, according to the physician. The mean rate of primary cesarean section (i.e., the rate for women without previous cesarean deliveries) was 17.2, with a range of 9.6 to 31.8 percent. A stepwise logistic-regression model of the determinants of primary cesarean section, including the individual physician, parity, birth weight, and maternal age and excluding specific medical indications, showed that only nulliparity (P less than 0.0001) was more important than the identity of the physician (P less than 0.001) in its influence on the rate of cesarean section. Variation in cesarean-section rates among physicians was not attributable to the practice setting, the patient population, the degree of obstetrical risk, or the physician's recent medicolegal experience, and it was not accompanied by corresponding differences in neonatal outcome. We conclude that individual practice style may be an important determinant of the wide variations in the rates of cesarean delivery among obstetricians. Our data do not permit us to say with certainty whether the procedure is overused by some obstetricians or underused by others, but we found no obvious differences in neonatal outcome associated with differences in the cesarean-section rate.