OBJECTIVES: We aimed to improve the nutritional care of preterm infants by developing a complex (multifaceted) intervention intended to translate current evidence into practice. We used the sociological framework of Normalization Process Theory (NPT), to guide implementation in order to embed the new practices into routine care. DESIGN: A prospective interventional study with a before and after methodology. PARTICIPANTS: Infants <30 weeks gestation or <1500 g at birth. SETTING: Tertiary neonatal intensive care unit. INTERVENTIONS: The intervention was introduced in phases: phase A (control period, January-August 2011); phase B (partial implementation; improved parenteral and enteral nutrition solutions, nutrition team, education, August-December 2011); phase C (full implementation; guidelines, screening tool, 'nurse champions', January-December 2012); phase D (postimplementation; January-June 2013). Bimonthly audits and staff NPT questionnaires were used to measure guideline compliance and 'normalisation', respectively. NPT Scores were used to guide implementation in real time. Data on nutrient intakes and growth were collected continuously. RESULTS: There were 52, 36, 75 and 35 infants in phases A, B, C and D, respectively. Mean guideline compliance exceeded 75% throughout the intervention period, peaking at 85%. Guideline compliance and NPT scores both increased over time, (r=0.92 and 0.15, p<0.03 for both), with a significant linear association between the two (r=0.21, p<0.01). There were significant improvements in daily protein intake and weight gain between birth and discharge in phases B and Ccompared with phase A (p<0.01 for all), which were sustained into phase D. CONCLUSIONS: NPT and audit results suggest that the intervention was rapidly incorporated into practice, with high guideline compliance and accompanying improvements in protein intake and weight gain. NPT appears to offer an effective way of implementing new practices such that they lead to sustained changes in care. Complex interventions based on current evidence can improve both practice and clinical outcomes.