Self-thinning-related mortality has been encountered in extensively cultured scallop populations. Here we report an experiment aimed at studying actual population dynamics of scallops undergoing self-thinning and develop a model of the process. Scallops were kept in small tanks at various initial population densities. Growth and survival were monitored for 39 months. Two cohorts were monitored (cohorts 1 and 2 were 1 and 2 years old, respectively). Growth was density-dependent in cohort 2. Survivorship was high in cohort 1 during the first summer but mass mortality occurred during the second summer. Mortality in cohort 2 peaked during the first summer. In both cohorts, peak mortality occurred at age 2+ years and thus was uncoupled among cohorts. Therefore, negative environmental factors are ruled out as explanations of mortality. Instead, it appears as if self-thinning interacted with some ontogenetic factor. Although cohort 1 was undergoing self-thinning, trends in biomass–density curves were non-monotonous because of mass mortality. In contrast, self-thinning in cohort 2 followed a classical pattern. One possible mechanism for such contrasting differences was that 1+-year-old scallops produced byssal threads and attached to one another during the first year, forming multilayered clumps. This behaviour disappeared during the second year and mass mortality was experienced as scallops adopted a single-layered arrangement. Although details may differ among species, the process depicted above provides a potential trigger to the sequence of events leading to mass mortality in scallop culture. Our model suggests that self-thinning occurred in all density groups, with a common slope but with different elevations in the biomass–density space.