Abstract Objectives Implementation intentions are ‘IF‐THEN’ plans that encourage goal‐intended behaviour. This study was designed to test whether an intervention encouraging the formation of implementation intentions can reduce self‐harm in the community. Design A randomized controlled design was used. Methods At pre‐intervention, outcome variables (self‐harm in both specified and unspecified critical situations and suicidality) and potential moderators of implementation intentions (goal intention, mental imagery, and exposure to self‐harm) were measured using self‐report questionnaires. The participants ( N = 469, aged 18–66 years, 86.4% female, 6.8% male and 6.7% other) were then randomized to either an experimental (implementation intention) or control task. At three‐months post‐intervention, self‐report questionnaires were used again to measure the outcome variables. Results There were no overall differences between the conditions at post‐intervention. However, goal intention and mental imagery, but not exposure to self‐harm, moderated the effects of condition on self‐harm in specified critical situations. At high (mean + 1 SD ) levels of both goal intention and mental imagery, the experimental condition reported self‐harming less frequently in the situations specified in their implementation intentions. Conclusions Implementation intentions therefore represent a useful intervention for reducing self‐harm in specified critical situations for people in the community who wish to avoid self‐harm and those who frequently experience self‐harm and suicide related mental imagery.