Postnatal growth failure is common in preterm infants, and one reason for this growth failure is that nutritional care is often variable and suboptimal. We aimed to improve the nutritional care of extremely preterm infants by implementing evidence-based practice changes and embedding these changes into routine care. Given the challenge that this represents, we used the novel sociological framework of normalisation process theory (NPT), to develop, guide, and assess the implementation process. We developed and implemented an evidence-based, complex intervention for the nutritional care of preterm infants, including guidelines, a screening tool, and a multidisciplinary ward round. We compared nutrient intakes and growth in infants born before 30 weeks' gestation or weighing less than 1500 g during 2012 (the intervention period, n=75) with those born in 2009 (pre-intervention period, n=65). The extent to which the new practices were being embedded into routine care was assessed with a bimonthly staff questionnaire based on NPT (n=80), allowing weaknesses to be detected and addressed and making implementation a dynamic process. Serial audits of guideline compliance were also done. Primary infant outcomes were daily energy and protein intake, and mean change in standard deviation scores (SDS) for weight and head circumference between birth and discharge. Linear regression was used to compare cohorts, with adjustment for gestational age and weight at birth. Mean daily energy and protein intakes as a percentage of recommended amounts improved significantly from 2009 to 2012 (energy 89·1% [SD 18·6] and protein 66·9% [20·5] vs 96·7 [18·4] and 78·2 [18·6]; p<0·05 for both). Growth also improved, with the mean change in weight SDS between birth and discharge changing from −0·90 (SD 0·78) to −0·57 (SD 0·75) (p<0·01). Head circumference did not improve significantly. During the intervention period, mean daily protein intake and mean NPT scores improved over time (r=0·88 [p=0·02] and 0·18 [p=0·01], respectively). There was a significant linear association between the mean audit score and the mean NPT score (r=0·22, p<0·01). The intervention succeeded in improving the nutrient intakes and growth of preterm infants. The NPT results show that these practices seem to have become part of routine care over time and suggest that measures of practice change using NPT can be related to measures of guideline compliance in the clinical setting. The findings of this study suggest that NPT offers an effective way of implementing and monitoring complex interventions. National Institute for Health Research.