To assess transvaginal (TV) and transabdominal (TA) cervical length (CL) measurements' variability and patient factors associated with TA CL accuracy. We hypothesized that patient factors would affect the accuracy of TA CL. This was a prospective cohort study. During anatomy ultrasound, TA and TV CL measurements were obtained, distance from placental edge to internal cervical os assessed, and demographic questionnaires completed. Patients between 18 to 22 weeks and 6 days were included and those <18 year old or with a twin gestation were excluded. TA CL >0.5 cm different from TV length was considered inaccurate. A total of 530 patients were included. Exactly 18.7% had a prior cesarean, 9.8% a preterm birth, and 2.2% a cervical procedure. Mean age and BMI were 31.1 years and 27.8 kg/m2 . Median number of living children was one. Median TA and TV CL were 3.42 and 3.53 cm. Exactly 36% (95% CI: 32-40%) of TA CL measurements were inaccurate. CL of 3.4 cm corresponded to a mean difference of zero between TA and TV CL. TA ultrasound had a sensitivity of 25% and a specificity of 98.5% to detect TV CL <2.5 cm. On multivariable analyses, Hispanic ethnicity was associated with inaccurate TA measurement (OR 0.48, 95% CI: 0.24-0.96, P = .04). On average, TA CL underestimates TV CL when TV CL >3.40 cm and overestimates TV CL when TV CL <3.40 cm. Additional co-variates did not impact accuracy. TA ultrasound has low sensitivity to predict short cervix. Relying solely on TA CL to identify those who need intervention may miss diagnoses. It may be reasonable to develop protocols in which TV CL is used for TA CL <3.4 cm.