OBJECTIVE-To determine if a policy of early amniotomy resulted in a reduction in mean labour duration when compared to a policy of conservation of the membranes. DESIGN-A single-centre randomized controlled trial. SETTING-A tertiary care teaching hospital in Alberta, Canada. SUBJECTS-Ninety-seven term nulliparae in spontaneous labour, baby in cephalic presentation. INTERVENTION-Early amniotomy versus intent to keep membranes intact. MAIN OUTCOME MEASURES-Interval from randomization to delivery, rate of abnormalities of fetal heart rate tracings, cord artery blood pH, Apgar scores. RESULTS-The mean interval from randomization to delivery was 390.9 (SE 29.1) min in the amniotomy group and 442.9 (SE 34.1) min in the control group (P = 0.251). There were no differences between groups in the occurrence of fetal heart rate tracing abnormalities, nor was there a difference in the proportion of babies with abnormal Apgar scores, or abnormal cord pH (less than 7.20). CONCLUSION-The results of the study fail to support the long held belief that early amniotomy is an effective method for reducing labour duration.