Objective: To examine the natural history of cervical length shortening in women who had experienced at least one prior spontaneous preterm birth at between 17 + 0 and 33 + 6 weeks' gestation. Methods: This was an analysis of prerandomization data from the multicenter Vaginal Ultrasound Cerclage Trial. Serial cervical length was measured by transvaginal sonography in 1014 high-risk women at 16 + 0 to 22 + 6 weeks. We performed survival analyses in which the outcome was cervical length shortening <25 mm and data were censored if this did not occur before 22 + 6 weeks' gestation. The incidence of cervical length shortening and the time to shortening were compared for women whose earliest prior preterm birth was in the mid-trimester, defined as <24 weeks, vs. those at weeks 24-33. Similar comparisons were performed based on each patient's most recent birth history. Results: Time to cervical length shortening by survival analysis was significantly shorter (hazard ratio (HR) = 2.2, P < 0.0001) and the relative risk (RR) of shortening significantly higher (RR = 1.8, P < 0.0001) for women whose earliest prior spontaneous preterm birth was at <24 weeks. A larger effect was observed for women whose most recent birth was at <24 weeks (HR = 2.8, P < 0.0001; RR = 2.1, P < 0.0001). The observed hazard ratios remained significant after adjusting for confounders in a multivariable Cox proportional hazards model. Conclusion: Women with a prior spontaneous preterm birth at <24 weeks are at a higher risk of cervical shortening, and do so at a higher rate and at an earlier gestational age, than do women with a later preterm birth history. Copyright © 2008 ISUOG. Published by John Wiley & Sons, Ltd.