© Queen’s Printer and Controller of HMSO 2017. Background: Behavioural counselling with intensive follow-up for obesity is effective, but in resource-constrained primary care settings briefer approaches are needed. Objectives: To estimate the clinical effectiveness and cost-effectiveness of an internet-based behavioural intervention with regular face-to-face or remote support in primary care, compared with brief advice. Design: Individually randomised three-arm parallel trial with health economic evaluation and nested qualitative interviews. Setting: Primary care general practices in the UK. Participants: Patients with a body mass index of ≥ 30 kg/m2 (or ≥ 28 kg/m2 with risk factors) identified from general practice records, recruited by postal invitation. Interventions: Positive Online Weight Reduction (POWeR+) is a 24-session, web-based weight management intervention completed over 6 months. Following online registration, the website randomly allocated participants using computer-generated random numbers to (1) the control intervention (n = 279), which had previously been demonstrated to be clinically effective (brief web-based information that minimised pressure to cut down foods, instead encouraging swaps to healthier choices and increasing fruit and vegetables, plus 6-monthly nurse weighing); (2) POWeR+F (n = 269), POWeR+ supplemented by face-to-face nurse support (up to seven contacts); or (3) POWeR+R (n = 270), POWeR+ supplemented by remote nurse support (up to five e-mails or brief telephone calls). Main outcome measures: The primary outcome was a modelled estimate of average weight reduction over 12 months, assessed blind to group where possible, using multiple imputation for missing data. The secondary outcome was the number of participants maintaining a 5% weight reduction at 12 months. Results: A total of 818 eligible individuals were randomised using computer-generated random numbers. Weight change, averaged over 12 months, was documented in 666 out of 818 participants (81%; control, n = 227; POWeR+F, n = 221; POWeR+R, n = 218). The control group maintained nearly 3 kg of weight loss per person (mean weight per person: baseline, 104.4 kg; 6 months, 101.9 kg; 12 months, 101.7 kg). Compared with the control group, the estimated additional weight reduction with POWeR+F was 1.5 kg [95% confidence interval (CI) 0.6 to 2.4 kg; p = 0.001] and with POWeR+R was 1.3 kg (95% CI 0.34 to 2.2 kg; p = 0.007). By 12 months the mean weight loss was not statistically significantly different between groups, but 20.8% of control participants, 29.2% of POWeR+F participants (risk ratio 1.56, 95% CI 0.96 to 2.51; p = 0.070) and 32.4% of POWeR+R participants (risk ratio 1.82, 95% CI 1.31 to 2.74; p = 0.004) maintained a clinically significant 5% weight reduction. The POWeR+R group had fewer individuals who reported doing another activity to help lose weight [control, 47.1% (64/136); POWeR+F, 37.2% (51/137); POWeR+R, 26.7% (40/150)]. The incremental cost to the health service per kilogram weight lost, compared with the control group, was £18 (95% CI-£129 to £195) for POWeR+F and-£25 (95% CI-£268 to £157) for POWeR+R. The probability of being cost-effective at a threshold of £100 per kilogram was 88% and 98% for POWeR+F and POWeR+R, respectively. POWeR+R was dominant compared with the control group. No harms were reported and participants using POWeR+ felt more enabled in managing their weight. The qualitative studies documented that POWeR+ was viewed positively by patients and that health-care professionals generally enjoyed supporting patients using POWeR+.