Objective. To examine the accuracy of abdominal palpation for identifying left-occipito-anterior (LOA) fetal position using abdominal ultrasound as the reference standard. Design: Classical test accuracy study undertaken in 2005–2007. Setting: Birmingham Women's Foundation NHS Trust serving a large, socio-economically and ethnically varied population. Sample. All nulliparous women with spontaneous or induced labor before established labor (cervix <4cm dilated), with a singleton live pregnancy of over 37 completed gestational weeks without known fetal abnormalities. Methods. Accuracy of abdominal palpation (index test) in identifying LOA fetal position, with abdominal ultrasound as reference. Trained observers blind to the index test results performed the ultrasound independently. Main outcome measures. Accuracy of palpation in determining LOA position. Results. Midwives’ abdominal palpation and ultrasound data were obtained from 629 women. There were 61 (9%) fetuses in LOA position that were verified by ultrasound. The sensitivity and specificity of palpation to detect LOA position were 34% (95%CI 23–46) and 71% (67–74), respectively. Midwives with experience >5 years achieved higher sensitivity compared to those with ≤5 years (odds ratio 4.02; 1.26–12.9; p=0.019). Sensitivity was higher for community compared with hospital midwives (OR 6.59; 1.11–39.11; p=0.038). Conclusions. Abdominal palpation to determine LOA position at the onset of labor had poor accuracy in nulliparous women on arrival at the maternity unit with a cervix dilation of <4cm. If future research demonstrates that an optimal fetal position at labor onset exists, ultrasound scan to confirm fetal position on arrival for birth may improve midwives’ ability to prognosticate.