OBJECTIVE: Shortening of the cervix that occurs early in gestation as determined by ultrasonography has been correlated with subsequent spontaneous preterm birth. Because previous studies have compared transperineal and endovaginal ultrasonographic assessment of the cervix over a wide gestational age range, we compared these methods in the midtrimester. STUDY DESIGN: We performed a prospective, blinded comparison of endovaginal and transperineal cervical ultrasonographic assessment. Unselected gravid women at 15-23 weeks' gestation were sequentially examined by 2 experienced sonographers using an endovaginal probe and a curvilinear probe. For each patient the initial sonographer and examination method were randomly assigned. The sonographer who performed the second examination was blinded to the results and images from the first examination. RESULTS: One hundred two women were studied at a mean of 19.6 weeks' gestation. The overall intermethod correlation was poor (R= 0.38). In 12 cases a transperineal measurement could not be obtained because of poor visualization of the required landmarks. In 33% of cases the intermethod difference in cervical length was ≥20%. There was no correlation in the identification of funneling at the internal os and very small correlation for the identification of a poorly developed lower uterine segment. CONCLUSION: Transperineal ultrasonographic imaging of the cervix is an unsatisfactory alternative to an endovaginal assessment in the midtrimester.