Objective To inform international research and policy, we conducted a meta-analysis of the experimental literature on pictorial cigarette pack warnings.

Data sources We systematically searched 7 computerised databases in April 2013 using several search terms. We also searched reference lists of relevant articles.

Study selection We included studies that used an experimental protocol to test cigarette pack warnings and reported data on both pictorial and text-only conditions. 37 studies with data on 48 independent samples (N=33 613) met criteria.

Data extraction and synthesis Two independent coders coded all study characteristics. Effect sizes were computed from data extracted from study reports and were combined using random effects meta-analytic procedures.

Results Pictorial warnings were more effective than text-only warnings for 12 of 17 effectiveness outcomes (all p<0.05). Relative to text-only warnings, pictorial warnings (1) attracted and held attention better; (2) garnered stronger cognitive and emotional reactions; (3) elicited more negative pack attitudes and negative smoking attitudes and (4) more effectively increased intentions to not start smoking and to quit smoking. Participants also perceived pictorial warnings as being more effective than text-only warnings across all 8 perceived effectiveness outcomes.

Conclusions The evidence from this international body of literature supports pictorial cigarette pack warnings as more effective than text-only warnings. Gaps in the literature include a lack of assessment of smoking behaviour and a dearth of theory-based research on how warnings exert their effects.