Promoting active travel can be beneficial for both health and the environment. However, evidence about the most effective strategies is inconsistent. We aimed to compare the effectiveness of interventions with positive (ie, carrot), negative (ie, stick), or a combination of strategies on changing population-level travel behaviour. We also aimed to identify which intervention functions, or mechanisms of how interventions seek to alter behaviour (eg, by addressing safety or accessibility), affect transport outcomes. For this systematic review and meta-analysis, we searched eight online databases for studies published before March 28, 2022: Web of Science, MEDLINE, Scopus, Applied Social Sciences Index and Abstracts, Global Health, PsycINFO, CINAHL, and Transport Research International Documentation. We did not restrict searches by language or publication date. We included controlled before-and-after studies of population-level interventions and travel behaviours (ie, driving, public transport, walking, and cycling) from adults in the general population. We categorised interventions according to their function. Depending on whether gains or losses due to intervention function could occur, we classified interventions as carrot (eg, new bike-share programmes), stick (eg, congestion charging), or combined carrot-and-stick interventions (eg, pedestrianising areas by use of reallocated parking space). We used harvest plots to summarise the findings and guide narrative synthesis. Where possible, we converted outcomes into standardised mean differences and did random-effects meta-analyses. From 38 916 records screened, 102 reports describing 121 interventions met the inclusion criteria. 79 interventions were carrots, 22 were carrot-and-sticks, and 20 were sticks. Results for carrot interventions were less consistent than for stick or combined interventions. Findings from the meta-analysis (64 reports describing 67 interventions) agreed with those in the narrative synthesis; although effects were statistically non-significant, for driving outcomes, interventions with stick strategies (standardised mean difference [SMD] -0·17, 95% CI -0·36 to 0·02) and combined carrot-and-stick strategies (-0·13, -0·47 to 0·20) had point estimates of greater magnitude than those for interventions with carrot strategies (-0·10, -0·23 to 0·03). Likewise, for active travel outcomes, combined carrot-and-stick strategies had a higher point estimate (0·33, -0·01 to 0·68) compared with carrot interventions (0·08, -0·05 to 0·21). Functions thought to change behaviour using financial means were effective at decreasing driving behaviour, whereas those improving access, safety, and space were effective for increasing active travel outcomes. This Article found that, although transport interventions with only positive strategies are more commonly evaluated, interventions that combine both positive and negative strategies might be more effective at encouraging alternatives to driving at the population level. Further research is needed for interventions involving a stick strategy, which remain less widely implemented or well studied than those with only carrot strategies. Medical Research Council, Cambridge Trust.