Previous literature on collusion has suggested that leniency programmes may have perverse effects. The reason behind this is that this type of mechanism seems to destabilise cartels, but they may have incentives to re-emerge quickly. In this paper, we experimentally investigate how the degree of leniency (full vs. partial) affects cartel stability and recidivism. We also examine the potential emergence of tacit collusion when communication is used by Competition Authorities to track and investigate firms. In a context with strong incentives for collusion and self-reporting of anticompetitive behaviour, we show that full leniency programmes are more effective in destabilising cartels because they induce more reporting than partial leniency programmes, but recidivism prevails as a major problem. Furthermore, costly communication reduces the frequency of communication but fails to reduce anticompetitive behaviour because it encourages tacit collusion between firms.