Lobster stock assessments are difficult because trap catch rates cannot be assumed to be proportional to abundance. A simulation model has been developed to consider the factors that might contribute to trap sampling bias. In the presence of trap saturation due to behavioural interactions at the trap, the model predicts an asymptotic relationship between catch and density and a decline in the variance : mean ratio of catch numbers per trap with density. A random or even distribution of lobsters among traps is predicted despite an initial aggregated distribution of lobsters on the seabed. Within the range of parameter values investigated, local depletion around the trap and decline in bait attractiveness had less influence than behavioural interactions on catch rates and the distribution of the catch among traps. All these results were, however, sensitive to assumptions made about the nature of lobster movement on the seabed. The results have implications for the use of catch data in assessing stock abundance, and as the pattern of model predictions is consistent with observed fisheries data, the model is likely to provide a useful framework for investigating capture processes in trap fisheries for crustaceans.