We investigate the desirability of excluding cartel ringleaders from leniency programs. Excluding ringleaders from both pre- and post-investigation leniency (full discrimination) has two opposite effects on cartel stability: a pro-competitive effect stemming from the asymmetry in payoffs induced by the differentiated leniency treatment, and a pro-collusive effect resulting from the reduction in the ringleader’s incentive to deviate. It is shown that excluding ringleaders from all types of leniency tends to stabilize cartels. We argue that a policy which denies post-investigation leniency to ringleaders while making them eligible for pre-investigation leniency eliminates the pro-collusive effect, and is superior to both full- and non-discrimination.