Electronic fetal monitoring (EFM) and cesarean sections (CS) have been studied in 14,484 patients delivered between 1970 and 1975. The primary CS rate for those monitored was 107/1000, and for the unmonitored was 57/1000. Multivariate analysis of these data shows that most of the increase in CS rate associated with monitoring is attributable to inherent differences between monitored and unmonitored patients. Furthermore, the association between monitoring and CSs was not homogeneous among sub-groups. Cesarean sections were relatively less frequent in monitored nulliparas with malpresentation and more frequent in otherwise normal monitored multiparas. These opposing effects tended to balance each other. During the study period the primary CS rate rose from 4.3 to 11.2%. This increase was strong in unmonitored patients and was primarily due to operations undertaken for fetopelvic dis-proportion. We cannot therefore attribute much of the increasing CS rate at our hospital to fetal monitoring.