Abstract Background To evaluate the self-reported pain scores as a predictor of preterm birth (PTB) in symptomatic twin pregnancy and to develop a nomogram for the prediction model. Methods We conducted a retrospective study of 148 cases of symptomatic twin pregnancies before 34 weeks of gestation visited at Seoul national university hospital from 2013 to 2018. With other clinical factors, self-reported pain score was evaluated by the numerical rating scale (NRS) pain scores for pain intensity. By multivariate analyses and logistic regression, we developed a prediction model for PTB within 7 days. Using the Cox proportional hazards model, the curves were plotted to show the predictability of the PTB according to NRS pain score, while adjusting the other covariates. Results Twenty-three patients (15.5 %) delivered preterm within 7 days. By a logistic regression analysis, higher NRS pain score (OR 1.558, 95 % CI 1.093–2.221, P < 0.05), shorter cervical length (OR 3.164, 95 % CI 1.262–7.936, P < 0.05) and positive fibronectin results (OR 8.799, 95 % CI 1.101–70.330, P < 0.05) affect PTB within 7 days. Using the variables, the area under the receiver operating characteristic curve (AUROC) of the prediction model was 0.917. In addition, we developed a nomogram for the prediction of PTB within 7 days. Conclusions Self-reported pain scores combined with cervical length and fetal fibronectin are useful in predicting impending PTB in symptomatic twin pregnancy.