BACKGROUND: One in five women experience psychological distress in the perinatal period. To support women appropriately, Australian guidelines recommend routine depression screening and psychosocial risk assessment by midwives in pregnancy. However, there is some evidence that current screening processes results in higher rates of false positives. The Perinatal Integrated Psychosocial Assessment (PIPA) Project compared two models of psychosocial assessment and referral - Usual Care and the PIPA model - with a view to improving referral decisions. This paper describes midwives' perspectives on psychosocial assessment, depression screening and referral at the antenatal booking appointment and compares midwives' experiences with, and perspectives on, the two models of care under investigation. METHODS: A two-phase, convergent mixed methods design was used. Midwives providing antenatal care completed a self-report survey in phase one prior to implementation of the new model of psychosocial assessment (n = 26) and again in phase two, following implementation (n = 27). Sixteen midwives also participated in two focus groups in phase two. Quantitative and qualitative data were compared and integrated in the presentation of results and interpretation of findings. RESULTS: Midwives supported psychosocial assessment believing it was a catalyst for 'Opening the door" to conversations with women. Midwives were comfortable asking the questions and tailored their approach to build rapport and trust. Overall. midwives expressed favourable views towards the PIPA model. A greater proportion of midwives relied mostly or entirely on the suggested wording for the psychosocial questions in the PIPA model compared to Usual Care (44.4% vs 12.0%, χ2=5.17, p=.023, φ =-.36). All midwives reported finding the referral or action message displayed at the end of the PIPA psychosocial assessment to be 'somewhat' or 'very' helpful, compared to 42.3% in Usual Care (χ2 = 18.36, p < .001, φ = -.64). Midwives were also more likely to act on or implement the message often or all of the time) in the PIPA model (PIPA = 69.2% vs Usual Care = 32.0%, (χ2 = 5.66, p < .017, φ = -.37). CONCLUSION: The study identified benefits of the new model and can inform improvements in psychosocial screening, referral and related care processes within maternity settings. The study demonstrates that psychosocial assessment can, over time, become normalised and embedded in practice.