Aims The aim of this paper was to explore responses to alcohol health information labels from a cross sectional survey of people who drink alcohol from 29 countries. Design This paper draws on findings from the Global Drug Survey (GDS) – an annual cross sectional online survey. Participants 75,969 (64.3% male) respondents from 29 countries were included in the study. Measures Respondents were shown seven health information labels (topics were heart disease, liver, cancer, calories, violence, taking two days off and myth of benefits of moderate drinking). They were asked if the information was new, believable, personally relevant, and if it would change their drinking. A multivariate multilevel Bayesian logistic regression model was used to estimate predicted probabilities for newness, believability, relevance and if messages would change drinking behaviour by country and information label. Findings Predicted probabilities showed substantial variability in responses across countries. Respondents from Colombia, Brazil and Mexico were more likely to consider drinking less as well as have lower levels of previous awareness. Those from Denmark and Switzerland were not as likely to say the labels would make them consider drinking less. The cancer message was consistently the newest and most likely to make people consider drinking less across countries. Conclusions Country differences in responses to messages can be used to create targeted harm reduction measures as well as inform what should be on labels. The provision of such health information on alcohol product labels may play a role in raising awareness of the risk of drinking. Global comparisons of responses to alcohol health information labels: a cross sectional study of people who drink alcohol from 29 countries.