Abstract When consumers have preference costs, two opposing effects need to be assessed to analyse the incentives of firms to set collusive prices. On the one hand, preference costs make a deviation from collusion less attractive, as the deviating firm must offer a large enough discount to cover the preference costs. On the other hand, preference costs lock in consumers and make punishment from rivals less effective. When preference costs are low, the latter of the two effects dominates and collusion is more challenging to sustain than in a situation with no preference costs. With high enough preference costs, collusion is a (weakly) dominant strategy. These results do not eventuate in a model with switching costs.