OBJECTIVES: to compare the impact of two methods of supplementary feeding of pre-term babies (bottle vs cup) on subsequent breast feeding and to assess the feasibility of using a randomised controlled trial (RCT) to investigate the topic. DESIGN AND METHOD: small scale prospective RCT. Data on breast feeding, as defined as the exclusive method of feeding, were collected. A range of relevant bio-data was also collected and their impact on breast feeding assessed. SETTING: a special care baby unit in a District General Hospital in the UK. PARTICIPANTS: over a three-month period, all pre-term babies (32-37 weeks' gestation) who fulfilled the inclusion criteria and has been born to mothers who had expressed a pre-partum desire to breast feed, who had consented to take part, were included (n=14). PROCEDURE: the eligible babies were randomly allocated to supplementary feeding of breast milk, via either a cup or a bottle. Whether or not the baby was being breast fed at discharge was noted. FINDINGS: the study suggested that this RCT framework is a viable method of investigating baby feeding. Because of the small-scale nature of the project, the actual database must be treated with extreme caution. No significant differences were found between the two groups in terms of breast feeding. However, the mothers reported high levels of support and also the breast-feeding rates were above the national averages. These two findings could have contributed to the non-significant results observed in this analysis. CONCLUSIONS: if the present findings could be supported by further research, then the non-significant results relating method of supplementary feeds to subsequent breast feeding could be explained by reference to three factors. Firstly, there is, in fact, no real effect of method of supplementary feeding and subsequent breast feeding; secondly, the method adopted differed from existing research and thus may be expected to produce non-corroborative results; and finally, the overall levels of breast feeding within the Unit generally were higher than the national average. The relevance of the RCT for investigating this subject is also discussed with reference to the present data set. Further experimental work to develop these ideas and to identify causal links is required.