We investigated the effect of an environmental perturbation on brown trout, Salmo trutta, dominance hierarchies. Hierarchies were established over a 1-week period under constant simulated natural conditions in artificial stream tanks. In the perturbation treatment water levels were then lowered for a week to simulate a drought, whereas conditions remained the same in the control tanks. We recorded behavioural interactions before and after the environmental perturbation. After the 2-week experiment, we killed the fish and measured growth rate, plasma cortisol, hepatic glycogen content, hepatosomatic index, gill epithelial chloride cell densities and interrenal cell nuclear areas. Aggression showed a nonsignificant increase in the drought tanks when the water level was lowered, and behaviour and social ranking of the fish were significantly affected by the environmental perturbation with a general breakdown in the social hierarchy. The pronounced benefits of dominance in terms of growth rate observed in the control tanks were not apparent in the drought tanks. However, the cortisol concentrations of the drought fish were not significantly higher than those of control fish at the end of the experiment, suggesting that the environmental change itself was not physiologically stressful in the long term. Neither were any other physiological parameters measured significantly different to those of the control tanks. Given that a stable social system (and its physiological consequences) was observed only in a constant environment, misleading conclusions may be drawn if environmental perturbations are not incorporated into experiments studying the behaviour of stream-living fish in simulated natural conditions.