Objective:To use meta‐analysis to evaluate the effect of epidural analgesia on the cesarean delivery rate.Data sources:The MEDLINE data base was searched for articles published in English between January 1981 and April 1992. We also interviewed experts and conducted a bibliographic follow‐up and manual review of recent journals published from April to July 1992.Methods of study selection:We excluded articles with irrelevant titles, and those case studies, book chapters, or articles that did not provide primary and relevant data. Two hundred thirty articles were read, including articles that reported on women of standard obstetric risk and on cesarean delivery rates for an epidural group and for a concurrent no‐epidural group. These criteria yielded six studies for a primary analysis and two others for a secondary analysis.Data extraction and synthesis:The sample size of the epidural and no‐epidural groups and the number of cesareans within each group were extracted. Tests of homogeneity were conducted. The pooled cesarean delivery risk difference as a result of epidural analgesia was estimated. The cesarean rate for women undergoing epidural analgesia was ten percentage points greater than for no‐epidural women (P< .05). More than a nine percentage point increase was shown for cesarean deliveries for dystocia (P< .05), when pooling either all studies or only randomized studies.Conclusions:The results of this meta‐analysis strongly support an increase in cesarean delivery associated with epidural analgesia. Further research should evaluate the balance between analgesia associated with the use of epidurals, and postpartum morbidity and costs associated with cesarean deliveries.(Obstet Gynecol 1994;83:1045‐52)