Objective To evaluate the predictive validity of the Chinese version of the schizophrenia early signs scale （ESS-C）. Methods A total of 70 patients with schizophrenia discharged from 6 mental health centers in Hongkou， Baoshan， Jiading， Songjiang， Changning and Xuhui districs of Shanghai from March to June 2018 were recruited. All patients were assessed by ESS-C and Positive and Negative Symptom Scale （PANSS） every 2 weeks until disease recurrence or follow-up for 6 months. The ESS-C scale scores of the recurrence group and the non-recurrence group were compared， and the predictive validity of the ESS-C was evaluated by constructing the curve of receiver operating characteristic （ROC）. Results Compared with the non-recurrence group， the scores of all dimensions and total scores of ESS-C in the last evaluation in the relapse group were higher， and the differences were statistically significant （P＜ 0.05）. The scale discriminative performance was good. The area under the ROC curve was 0.886， 95%CI valued from 0.803 to 0.970. The sensitivity， specificity， positive predictive value and negative predictive value were 78.6%， 89.3%， 64.7% and 94.3%， respectively， with 21.5 points as the optimal cut-off value for recurrence. Conclusions The ESS-C has good predictive accuracy and is a simple and reliable tool for evaluating the relapse of schizophrenia in the community.