目的：初步探讨国外文献报道的7种预测模型在国内临床中对剖宫产术后经阴道分娩（vaginal birth after cesarean，VBAC）的预测价值。方法：收集2007年1月1日至2014年11月31日共53例剖宫产术后再次妊娠阴道试产（trial of labor after cesarean，TOLAC）产妇的相关资料及其临床结局，采用Troyer、Weinstein、Flamm、Gonen、Smith、Grobman、Torri所建立的预测模型对每例患者进行评分，t检验比较应用每种模型预测VBAC失败组与成功组的分值。对成功组及失败组的评分进行分层，采用秩和检验比较每种评分对于成功率预测的相关性；应用受试者工作特征（receiver operating characteristic，ROC）曲线及曲线下面积（area under curve，AUC）评价预测模型的分辨力。结果：53例产妇中，VBAC成功者44例，失败者9例，成功率83.0%。成功组及失败组的分值在Weinstein及Grobman预测模型中差异有统计学意义，在其他各预测模型无明显差异。评分经分层后，VBAC的成功率在各层与原文献中各模型的预测相符，各层间的成功率差异仅在Weinstein预测模型中存在统计学差异。Weinstein预测模型的ROC曲线AUC最大（0.746），但各预测模型间无显著差异。结论：7种预测模型中，Weinstein对本研究人群VBAC成功率的预测优于其他6个模型，但仍需建立符合我国具体情况的VBAC预测模型。

Objective：To evaluate the seven existing vaginal birth after cesarean (VBAC) screening tools and to identify additional factors that may predict VBAC or failed trial of labor in China. Methods: In the study, 53 patients with 1 previous cesarean delivery who then delivered between January 1, 2007 and Novenber 31, 2014 were recruited. The average age of the patients was (32.1±3.5) years，the average gestational age was (38.0±2.3) weeks. There was no significant difference of the successful group and the failed group in the maternal/neonatal mortality and morbidity, also in the incidence of the postpartum hemorrhage and the postpartum infection. The probability of VBAC was calculated for each participant using 7 prediction models created by Weinstein, Flamm, Grobman, Gonen, Troyer, Smith and Torri. The data were analyzed using t test, rank-sum test, and receiver operating curve analysis. Results: 44 trial of labor patients had a vaginal birth after cesarean delivery, and the successful rate was 83%. The scores between the successful group and the failed group had significant difference when eva-luated by Weinstein and Grobman scoring models only. After recalculating the successful rate of VBAC in different score levels according to the references, there was significant difference between the rates of different score levels when evaluated by the Weinstein model. The successful rates of different score levels were higher compared to the references (<50%) when evaluated by the Troyer (70%), Gonen (60%), Torri (85.7%) models. The area under the receiver operating characteristic curve of Weinstein prediction model (0.746) and Flamm prediction model (0.723) were more than 0.7, and there was no significant difference between the seven models. Conclusion: Among the seven scoring models, the Weinstein model is more applicable to the population of our country, but a new model more applying to Chinese women still needs to be created.