Around one in five children in England is obese when they leave primary school. Thus far, it has not been demonstrated that primary care interventions to manage childhood obesity can achieve significant weight reduction. Training obese children to eat more slowly as an adjunct to other healthy lifestyle behaviour change has been shown to increase weight reduction in a hospital setting. This pilot study aimed to test recruitment strategies, treatment adherence, clinic attendance and participants' experiences of using a device [Mandolean® (previously Mandometer®, Mikrodidakt AB, Lund, Sweden)] to slow down speed of eating as an adjunct to dietary and activity advice in treating obesity in primary school-aged children. A two-arm, parallel, randomised controlled trial with a qualitative study embedded within the pilot. Randomisation occurred after informed consent and baseline measures were collected. Participants were randomised by the Bristol Randomised Trials Collaboration randomisation service with allocation stratified by hub and minimised by age of the child, gender, and baseline body mass index (BMI) standard deviation score (BMI z-value) of the child, and by BMI of the study parent (obese/not obese). General practices across Bristol, North Somerset and South Gloucestershire primary care trusts. Children (BMI ≥ 95th percentile) aged 5-11 years and their families. Standard care comprised dietary and activity advice by trained practice nurses. Adjunctive Mandolean training (the intervention) educated participants to eat meals more slowly and to rate levels of fullness (satiety). Mandolean is a small computer device attached to a weighing scale that provides visual and oral feedback during meals while generating a visual representation of levels of satiety during the meal. Participants were encouraged to eat their main meal each day from the Mandolean. One parent was also given a Mandolean to use when eating with the child. Outcomes for the pilot were recruitment of 36 families to the trial in the 9-month pilot phase, that meals would be eaten at least five times a week off a Mandolean by 90% of patients randomised to the intervention arm, that 80% of patients in both arms would attend the weight management clinic appointment 3 months post randomisation and that > 60% of children using Mandolean would demonstrate a reduction in speed of eating from baseline within 3 months of randomisation. None of the criteria for progression to the main trial were reached. Despite numerous pathways being available for referral, only 21 (13 to standard care, eight to intervention arm; 58%) of the target 36 families were recruited in the pilot phase. Less than 20% of those randomised to Mandolean used the device at least five times a week. The > 60% target for slowing down of eating speed by 3 months was unmet. Attendance at the weight management clinic in general practice hubs for both arms of the study at 3 months was 44% against a target of 80%. This pilot trial failed to meet its objectives in terms of recruitment, treatment adherence, demonstration of a reduction in speed of eating in sufficient numbers of children, and attendance at follow-up appointments. Despite a high prevalence of childhood obesity in the geographical area and practices signing up for the trial, this study, like many others, demonstrates a failure of families to engage with and respond to primary care weight management interventions. We need to understand why the target population seems inured to the health message that childhood obesity is a significant health-care issue and identify the barriers to seeking help and then acting on positive health behaviour retraining. Only when we have fully understood the general public's perceptions of childhood obesity and have identified ways of engaging target populations can we hope to develop interventions that can work in a primary or community-based setting. Current Controlled Trials ISRCTN90561114. This project was funded by the NIHR Health Technology Assessment programme and will be published in full in Health Technology Assessment; Vol. 18, No. 47. See the NIHR Journals Library website for further project information.