Estimates of average cartel duration and the annual probability of cartel death are based on data for discovered cartels. It is recognised that these estimates could be biased because the population of discovered cartels may not be a representative sample of the latent population of cartels. This article constructs a birth-death-discovery process to investigate the source and direction of possible biases. Bayesian inference is used to provide bounds on the extent of the bias and deliver an improved set of beliefs on the probability of cartel death.