OBJECTIVE: To investigate whether serum relaxin level is associated with preterm birth in symptomatic women, either as a standalone test or in the context of a combined model of serum relaxin and cervical length (CL).METHODS: This was a case-control study of women with a singleton pregnancy who presented between 24 + 0 and 26 + 6 weeks' gestation with threatened preterm labor and intact membranes. CL, full blood count, C-reactive protein level and maternal demographics were recorded at presentation, and blood samples were taken for relaxin measurement. Parameters were compared between women who delivered preterm (before 37 weeks) (n = 46) and those delivering at term (n = 66). Logistic regression with receiver-operating characteristics (ROC) curve analysis was used to assess significant predictors for birth before 37 and before 34 weeks.RESULTS: Women delivering before 37 weeks had higher mean serum relaxin levels and lower mean CL than those delivering at term (P < 0.0001). Relaxin alone had 63% (95% CI, 49-75%) sensitivity for birth before 37 weeks and 61% (95% CI, 47-74%) for birth before 34 weeks, at a 10% false-positive rate (FPR). Serum relaxin levels did not correlate with CL; a combined model of the two predictors had an area under the ROC curve of 0.895 (95%CI, 0.835-0.954) for the prediction of birth before 37 weeks and 0.869 (95% CI, 0.802-0.937) for birth before 34 weeks (n =  44). Serum relaxin > 1010 pg/mL had 58% sensitivity for prediction of preterm birth in women with a CL > 15 mm, at a 10% FPR.CONCLUSIONS: High serum relaxin level is associated with an increased risk of preterm birth in second-trimester symptomatic women with intact membranes. A combination of serum relaxin and CL increases predictive accuracy for preterm birth. Copyright © 2017 ISUOG. Published by John Wiley & Sons Ltd.