The objective of this study was to investigate the accuracy of universal third trimester umbilical artery (UA) Doppler to predict adverse pregnancy outcome at term. We searched Medline, EMBASE, the Cochrane library and ClinicalTrials.gov from inception to October 2020 and we also analyzed previously unpublished data from a prospective cohort study of nulliparous women, the Pregnancy Outcome Prediction (POP) study. We included studies that performed a third-trimester ultrasound scan in unselected, low or mixed risk populations, excluding studies which only included high risk pregnancies. Meta-analysis was performed using the hierarchal summary receiver operating characteristic curve (HSROC) analysis and bivariate logit-normal models. We identified 13 studies (including the POP study) involving 67,764 pregnancies which met our inclusion criteria. The overall quality was variable and only six studies (N = 5777 patients) blinded clinicians to the UA Doppler result. The summary sensitivity and positive likelihood ratio (LR) for small for gestational age (SGA; birthweight <10th centile) were 21.7% (95% CI 13.2-33.6%) and 2.65 (95% CI 1.89-3.72) respectively. The summary positive LR for NICU admission and metabolic acidosis were 1.35 (95% CI 0.93-1.97) and 1.34 (95% CI 0.86-2.08) respectively. The results were similar in the POP study: associations with SGA (positive LR 2.66 [95% CI 2.11-3.36]) and severe SGA (birthweight <3rd centile; positive LR 3.27 [95% CI 2.29-4.68]) but no statistically significant association with neonatal morbidity. We conclude that third trimester UA Doppler has moderate predictive accuracy for small for gestational age but not for indicators of neonatal morbidity in unselected and low risk pregnancies.