BackgroundA perception that caesarean delivery is safer than vaginal delivery might be causing China's high rates of caesarean delivery. The aim of this study was to analyse maternal and perinatal outcomes by mode of delivery to elucidate the relative short-term safety of delivery choices. MethodsWe did a cross-sectional analysis of all deliveries from 2011 from 39 hospitals in three regions of Mainland China. Univariate and multilevel analysis were performed to study the association between the mode of delivery and the maternal and neonatal outcome. In the multivariate analysis, risks of maternal and neonatal outcomes associated with delivery outcomes were presented by adjusted odds ratios (OR) with corresponding 95% CIs. Maternal outcomes included postpartum haemorrhage, maternal death, combined severe outcomes (maternal death, transfusion, hysterectomy), and perinatal outcomes included neonatal death, perinatal death, ICU admission, 5 min Apgar less than 4, respiratory distress syndrome, infection, hypoxic ischaemic encephalopathy, birth trauma, and meconium aspiration. Ethical approval was not required with use of this pre-existing de-identified database. FindingsOf 109 806 deliveries, 59 561 (54%) were by caesarean delivery, 19 428 (18%) were antepartum non-indicated, 19 339 (18%) were antepartum indicated, 3251 (3%) were intrapartum non-indicated, and 17 543 (16%) were intrapartum indicated. 22 679 (38%) of caesarean deliveries were non-indicated, including 13 792 (23%) caesareans on maternal request, and 8887 (16%) caesareans for antepartum cephalo-pelvic disproportion, maternal age of more than 35 years, myopia, isolated nuchal cord, or precious baby (defined as in-vitro pregnancy or poor obstetric history such as prior fetal death, neonatal death, or anomaly). Compared with spontaneous vaginal delivery, antepartum non-indicated caesarean delivery had lower rates of postpartum haemorrhage (OR 0·76 [95% CI 0·68–0·75]), and we found no significant difference in rates of maternal death or combined severe outcomes (death, transfusion, hysterectomy). Labours resulting in intrapartum caesarean delivery had a higher rate of postpartum haemorrhage than vaginal delivery (1·6 [1·46–1·75]). Compared with spontaneous vaginal delivery, antepartum non-indicated caesarean delivery had low rates of neonatal death (0·21 [0·11–0·41]), perinatal death (0·15 [0·09–0·26]), admission to ICU (0·54 [0·42–0·69]), 5 min apgar of less than four (0·18 [0·9–0·36]), and respiratory distress syndrome (0·36, 0·22–0·59). Rates of infection, hypoxic ischaemic encephalopathy, birth trauma, or meconium aspiration did not differ between births by spontaneous vaginal delivery and antepartum non-indicated caesarean delivery. Labours resulting in intrapartum caesarean delivery had low rates of neonatal death (0·15 [0·09–0·28]), perinatal death (0·08 [0·05–0·11]), and 5 min apgar of less than four (0·13 [0·1–0·19]) but increased rates of admission to ICU (1·29 [1·10–1·52]), respiratory distress syndrome (1·42 [1·11–1·82]), infection (2·33 [1·32–4·10]), and meconium aspiration (2·23 [1·3–3·09]) than delivery by spontaneous vaginal delivery. InterpretationScheduled non-indicated caesarean delivery was associated with short-term maternal and perinatal outcomes that were as safe or safer than scheduled vaginal delivery or labour resulting in caesarean delivery. This safety profile might change as more repeat cesarean deliveries are performed given the liberalisation of the one-child policy in 2013, allowing more women to have a second pregnancy. FundingHealth industry special funds for Public Benefit Research Foundation from the Ministry of Health, China (grant number 201002013).