Background: In the UK, almost all maternity care (>99% of births) is delivered via the National Health Service, which serves a varied population (approximately 22% from ethnic minority groups) according to a set of agreed standards and guidelines with common training pathways for maternity professionals. This makes the UK a useful high-income context in which to investigate maternity care and outcomes. Increasing availability of electronic health record data for women giving birth has made it possible to understand risk factors for adverse outcomes and the impacts of policy change in maternity care more closely. Furthermore, there is growing attention to ethnic and socioeconomic inequalities in outcomes of maternity care. The overall aim of this thesis is to demonstrate how data collected during maternity care can be used to understand determinants of maternity outcomes. In particular, I look at the association between women’s socioeconomic and ethnic background and their maternity outcomes. Methods: In this thesis, observational epidemiological studies using national patient-level datasets address four related issues in maternity care in England and Wales. First, the quality of coding of ethnicity is evaluated in a cross-validation study comparing two sources of ethnicity data for women giving birth. Second, risk factors for adverse pregnancy outcomes (postpartum haemorrhage, maternal intensive care admission, and preterm birth) are examined using multivariable logistic regression models adjusting for clinical risk factors and care received. Third, the performance of the risk-classification system used to determine women’s choice of birthplace, using the National Institute for Care Excellence guideline for Intrapartum Care, is evaluated by calculating the proportion of women in each risk group who experience a complicated birth requiring obstetric or neonatal assistance. Fourth, the proportion of adverse pregnancy outcomes (stillbirth, preterm birth, and fetal growth restriction) attributable to socioeconomic and ethnic inequality is estimated using population attributable fractions. Results: First, cross-validation of ethnicity data between datasets supports the use of ethnicity collapsed into groups, with caution over results for women with mixed ethnicity, for whom the most inconsistencies are observed. Second, studies examining risk factors for severe maternal morbidity (maternal intensive care admission and postpartum haemorrhage) demonstrate evidence that these outcomes are more common for Black women than women from other ethnic groups; this association persists following adjustment for clinical characteristics and differences in care given. Furthermore, detailed evaluation of risk factors for preterm birth demonstrates that different groups of women experience iatrogenic (provider-initiated) and spontaneous preterm birth, and these should be measured separately. Third, giving more weight to parity and history of previous caesarean improves the risk assessment of women giving birth at term in comparison to currently used classification methods. Finally, ethnic and socioeconomic inequalities are responsible for a substantial proportion of stillbirths, preterm births and babies born with fetal growth restriction; while socioeconomic inequalities are partially attenuated by adjustment for the modifiable maternal risk factors BMI and smoking, ethnic inequalities are not. Conclusions: Increasing availability of clinical data have made it possible to evaluate maternity care in more depth, demonstrating lessons for clinical risk assessment and care, avenues for further research development, and potential targets for political and public health interventions to improve the health and circumstances of women before and during pregnancy. As electronic records become more widespread and comprehensive, the quantity and sophistication of questions it will be possible to answer will expand, encompassing a wider reach of women’s healthcare before, during and after birth. This thesis demonstrates that such data, if handled carefully, can support our understanding of individual risk factors, risk classification, and healthcare systems and policy, and be used to develop recommendations to improve both healthcare policy and clinical care for women and their families.