Mental imagery interventions are a cost-effective way of promoting health behaviour change. We tested a mental imagery intervention designed to promote adherence to wearing face coverings during the COVID-19 pandemic. A four-arm randomised controlled trial to explore potential mechanisms of action. Measures of behaviour (frequency of self-reported face covering adherence), theory of planned behaviour constructs (i.e. intention, attitudes, subjective norms, and perceived behavioural control), personality traits, imagery ability and barrier self-efficacy were measured at baseline (T1). Behaviour was also assessed at four-week follow-up (T2). Of 297 participants, a majority always wore face coverings ( 216, 73% overall sample). Logistic regression analyses revealed no intervention effects on changes in adherence to wearing face coverings, though T1 wearing of face coverings and being male predicted T2 behaviour. Subgroup analysis of participants self-reporting 'suboptimal T1 adherence, revealed that T2 non-adherence was predicted by being a non-student and by subjective norms and lower T1 intention to wear face coverings. Imagery-based interventions to increase face covering wearing adherence may exert significant public health effects but only when conducted on a very large scale. Our findings suggest that interventions should target men and disrupt habitual past behaviour.