Objective To investigate the value of PET/CT biphasic imaging in predicting lymph node metastasis of cervical carcinoma. Methods We retrospectively analyzed 202 lymph nodes from January 2019 to July 2021 in 97 patients with cervical carcinoma who underwent surgery and had preoperative PET/CT for early and delayed imaging (duplex). The metabolic parameters of lymph nodes in early imaging, such as SUVmaxL1, SUVmean, SUVpeak, MTV, TLG, length-short diameter ratio (L/D), SUVmaxT of lesions, SUVmaxA of abdominal aorta, SUVmaxH of liver, and SUVmaxL2 of lymph nodes in delayed imaging were measured. The corresponding ratios, △SUVmax (SUVmaxL2-SUVmaxL1) and retention index (RI) were calculated. The differences of each parameter between lymph node metastasis group(n=42) and non-metastasis group(n=160) were compared. Multivariate regression analysis was performed in combination with traditional diagnostic criteria to establish a combined predictive diagnostic model. Receiver operating characteristic (ROC) curve was used to compare the diagnostic efficacy. Results SUVmaxL1, SUVmean, SUVpeak, MTV, TLG, L/D, SUVmaxL2, △SUVmax, RI, SUVmaxL1/T, SUVmaxL1/A and SUVmaxL1/H were significantly different between metastasis and non-metastasis groups, and Logistic multi-factor regression analysis showed that SUVpeak, L/D, conventional PET/CT diagnostic criteria and SUVmaxL2 were independent risk factors for lymph node metastasis of cervical cancer; the area under curve (AUC) of conventional PET/CT diagnostic criteria was 0.839, with a sensitivity of 67.7 % and specificity of 87.5%; the AUC of single-temporal combined predictive diagnostic model was 0.915, with a sensitivity and specificity of 83.3% and 85.6%, respectively; the AUC of dual'temporal combined predictive diagnostic model was 0.995, with a sensitivity and specificity of 97.6% and 98.7%, respectively. The AUC of the single-temporal combined predictive diagnostic model was higher than that of the conventional PET/CT diagnostic criteria, the AUC of the dual-temporal combined predictive diagnostic model was higher than that of the conventional PET/CT diagnostic criteria, and the AUC of the dual-temporal combined predictive diagnostic model was higher than that of the single-temporal image.The difference was statistically significant (P<0.05). Conclusion The diagnostic efficacy of single-temporal and dual'temporal combined predictive diagnostic models for lymph node metastasis is higher than that of conventional diagnostic criteria, and the diagnostic efficacy of dual-temporal combined predictive diagnostic model is significantly higher than that of single-temporal combined predictive diagnostic model.