A systematic review, meta-analysis and meta-regression were conducted to evaluate the effectiveness of behavioural weight management programmes and examine how programme characteristics affect mean weight loss. Randomized controlled trials of multicomponent behavioural weight management programmes in overweight and obese adults were included. References were obtained through systematic searches of electronic databases (conducted November 2012), screening reference lists and contacting experts. Two reviewers extracted data and evaluated risk of bias. Thirty-seven studies, representing over 16,000 participants, were included. The pooled mean difference in weight loss at 12 months was −2.8 kg (95% confidence interval [CI] −3.6 to −2.1, P < 0.001). I2 indicated that 93% of the variability in outcome was due to differences in programme effectiveness. Meta-analysis showed no evidence that supervised physical activity sessions (mean difference 1.1 kg, 95% CI −2.65 to 4.79, P = 0.08), more frequent contact (mean difference −0.3 kg, 95% CI −0.7 to 0.2, P = 0.25) or in-person contact (mean difference 0.0 kg, 95% CI −1.8 to 1.8, P = 0.06) were related to programme effectiveness at 12 months. In meta-regression, calorie counting (−3.3 kg, 95% CI −4.6 to −2.0, P = 0.027), contact with a dietitian (−1.5 kg, 95% CI −2.9 to −0.2, P < 0.001) and use of behaviour change techniques that compare participants' behaviour with others (−1.5 kg, 95% CI −2.9 to −0.1, P = 0.032) were associated with greater weight loss. There was no evidence that other programme characteristics were associated with programme effectiveness. Most but not all behavioural weight management programmes are effective. Programmes that support participants to count calories or include a dietitian may be more effective, but the programme characteristics explaining success are mainly unknown.