Antenatal mental health assessment is increasingly common in high-income countries. Despite lacking evidence on validation or acceptability, the Whooley questions (modified PHQ-2) and Arroll ‘help’ question are used in the UK at booking (the first formal antenatal appointment) to identify possible cases of depression. This study investigated validation of the questions and women’s views on assessment. Women (n = 191) booking at an inner-city hospital completed the Whooley and Arroll questions as part of their routine clinical care then completed a research questionnaire containing the Edinburgh postnatal depression scale (EPDS). A purposive subsample (n = 22) were subsequently interviewed. The Whooley questions ‘missed’ half the possible cases identified using the EPDS (EPDS threshold ≥10: sensitivity 45.7 %, specificity 92.1 %; ≥13: sensitivity 47.8 %, specificity 86.1 %), worsening to nine in ten when adopting the Arroll item (EPDS ≥10: sensitivity 9.1 %, specificity 98.2 %; ≥13: sensitivity 9.5 %, specificity 97.1 %). Women’s accounts indicated that under-disclosure relates to the context of assessment and perceived relevance of depression to maternity services. Depression symptoms are under-identified in current local practice. While validated tools are needed that can be readily applied in routine maternity care, psychometric properties will be influenced by the context of disclosure when implemented in practice.