Objective: The purpose of this study was to compare second-trimester transvaginal cervical cerclage with conservative management on duration of pregnancy and perinatal outcome in patients with early or advanced cervical changes. Study Design: A historical cohort analysis was performed. Maternal and neonatal records between 1995 and 1999 were retrospectively reviewed for women presenting between 18 and 27 weeks of gestation with early cervical changes (length <3 cm, dilatation <2 cm, funneling of fetal membranes shown by transvaginal ultrasonography) (group 1, n = 31) and for women with advanced cervical effacement and dilatation (cervical dilatation ≥2 cm but ≤5 cm, fetal membranes visible) (group 2, n = 39). In each group, patients who underwent Shirodkar or McDonald cerclage were compared with patients treated conservatively with bed rest. Both groups also received multifactorial treatment with tocolytic agents, broad-spectrum antibiotics, and indomethacin. Outcome variables were analyzed for statistical significance by parametric and nonparametric methods. Results: Regardless of treatment method, patients with early cervical changes (group 1) were given a diagnosis earlier and delivered later in pregnancy compared with their counterparts who had advanced cervical changes (group 2) ( P < .05). In both patients who underwent cerclage and those treated conservatively, the mean birth weight among surviving infants was higher and the mean neonatal intensive care unit stay was shorter in group 1 than in group 2 ( P < .02). However, duration of maternal hospital stay and neonatal survival rates were not different. In both groups 1 and 2, the interval from treatment to delivery, the mean gestational age at delivery, and mean birth weight were increased, whereas neonatal intensive care unit stay was decreased by cerclage treatment ( P < .05). In group 1, a higher percentage of patients treated with cerclage received antibiotics and indomethacin than did control subjects ( P < .01), whereas in group 2, the use of multifactorial treatment was not different ( P = .5). The duration of maternal hospital stay and neonatal survival did not differ significantly among patients treated conservatively or with cerclage. Conclusions: Diagnosis of premature cervical changes by ultrasonography was correlated with treatment earlier in gestation and with a favorable impact on perinatal outcome in both patients treated with cerclage and those treated conservatively. Cervical cerclage was associated with an improved perinatal outcome (in comparison with conservative therapy) in women with early cervical changes detected by ultrasonography and in patients with advanced cervical dilatation and visible membranes. However, the apparent therapeutic effect of cerclage in patients with mild cervical incompetence may be due in part to an increased use of antibiotics and indomethacin in conjunction with cerclage. (Am J Obstet Gynecol 2001;184:1447-56.)