Reliable estimates of crime detection probabilities could help in designing better sanctions and improve our understanding of the efficiency of law enforcement. For cartels, we only have limited knowledge on the rate at which these illegal practices are discovered. In comparison to previous works, this paper offers a more parsimonious and simple-to-use method to estimate time-dependent cartel discovery rates, while allowing for heterogeneity across firms. It draws on capture–recapture methods that are frequently used in ecology to make inferences on various wildlife population characteristics. An application of this method provides evidence that less than a fifth of cartelising firms are discovered. Copyright © 2013 John Wiley & Sons, Ltd.