Hostility towards outgroups contributes to costly intergroup conflict. Here we test an intervention to reduce hostility towards Muslims, a frequently targeted outgroup. Our 'collective blame hypocrisy' intervention highlights the hypocrisy involved in the tendency for people to collectively blame outgroup but not ingroup members for blameworthy actions of individual group members. Using both within-subject and between-subject comparisons in a preregistered longitudinal study in Spain, we find that our intervention reduces collective blame of Muslims and downstream anti-Muslim sentiments relative to a matched control condition and that the effects of the intervention persist one month and also one year later. We replicate the benefits of the intervention in a second study. The effects are mediated by reductions in collective blame and moderated by individual differences in preference for consistency. Together, these data illustrate that the collective blame hypocrisy intervention enduringly reduces harmful intergroup attitudes associated with conflict escalation, particularly among those who value consistency in themselves and others. An intervention highlighting the hypocrisy involved in blaming Muslims but not ingroup members reduces collective blame of Muslims. This reduction was maintained for a year, despite an attack by Muslim extremists that was carried out during this time.