Dense blue mussel assemblages are unstable, their structure changing from year to year. Three types of models can be used to describe this instability: (1) “exogenous” model based on regional temperature fluctuations, (2) “endogenous” deterministic model associated with negative impact of adult mussels on juveniles and (3) “density-linked stochasticity” model based on positive feedbacks resulting in overcrowding and destabilizing the settlement. We compared predictions deduced from these models with a time series based on the results of long-term (18 years) monitoring of abundance and demographic structure of three mussel beds at the White Sea. Most of our findings agreed well with the predictions deduced from the endogenous model. In particular, (1) long-term changes in mussel abundance and demographic structure were strictly cyclic, with non-matching periods (5–9 years) at different sites; (2) stages with the dominance of old mussels alternated with those where juveniles dominated and (3) some signals of de