Background
Using a fixed cut‐off of ≤25 mm, ultrasound assessment of cervical length during the 18–23 week anomaly scan has been shown to identify approximately 50% of pregnancies that would deliver prior to 34 weeks.

Aim
To determine whether a policy of reverting to transvaginal cervical assessment only if the cervix appears short (≤25 mm) on transabdominal assessment affects the efficiency of screening.

Methods
Women with a singleton pregnancy that presented for a routine anomaly scan had their cervical length assessed transabdominally, initially with the maternal bladder full (TABF) and then empty (TABE). Cervical length was then assessed transvaginally (TV).

Results
One hundred and ninety‐eight women agreed to participate in the study. Identification of the internal and external cervical os was possible during TABF, TABE and TV sonography in 97.0, 82.8 and 100%, respectively. Compared with TV sonography, TABF overestimates cervical length (6.1 mm difference in median values; P < 0.01). There was no significant difference between TV and TABE. However, TABE assessment was not possible in one in six women. If TABF sonography was to be used as a screening tool and using ≤25 mm as the critical cut‐off, the sensitivity and specificity was 15.4 and 93.2%, respectively. Conclusion This study has shown that assessment of cervical length using a TA approach is only routinely possible when the bladder is full. However, measurements are significantly overestimated. Therefore, we feel that TV assessment of cervical length is the preferred method of reliable cervical assessment. As such, all women should be offered a TV assessment of cervical length at the time of the fetal anomaly ultrasound as a screening test for preterm birth.