Objectives We aimed to investigate which prevention strategies for low back pain (LBP) are most effective. Design We completed a Bayesian network meta-analysis to summarise the comparative effectiveness of LBP prevention strategies. The primary outcomes were an episode of LBP and LBP-associated work absenteeism represented as ORs with associated 95% credibility intervals (CrIs). We ranked all prevention strategies with surface under the cumulative ranking curve (SUCRA) analysis. Data sources PubMed, EMBASE and CENTRAL databases were searched along with manual searches of retrieved articles. We only included randomised controlled trials (RCTs) that reported an episode of LBP and/or LBP-associated work absenteeism evaluating LBP prevention strategies were included. Eligibility criteria for selecting studies Data were independently extracted by two investigators, and RCT quality was assessed using the Cochrane Risk of Bias tool. Results and summary Forty RCTs were included. Exercise combined with education (OR: 0.59, CrI: 0.41 to 0.82) and exercise alone (OR: 0.59, CrI: 0.36 to 0.92) both prevented LBP episodes; exercise combined with education and education alone both had large areas under the curve (SUCRA: 81.3 and 79.4, respectively). Additionally, exercise (OR: 0.04, CrI: 0.00 to 0.34) prevented LBP-associated work absenteeism, with exercise and the combination of exercise and education ranking highest (SUCRA: 99.0 and 60.2, respectively). Conclusions Exercise alone and exercise combined with education can prevent episodes of LBP and LBP-related absenteeism.