Much of the knowledge on salmonid ecology is based on electrofishing. Because there is little information on how electrofishing affects the subsequent behavior of fish, some of the salmonid data may be unreliable. Estimates of population sizes of stream fish based on mark–recapture or removal rely to a large extent on electrofishing as well. The population estimates obtained can be affected by high rates of movement suggested among stream-dwelling fish, regardless of whether the movements are natural or sampling induced. The amount of movement before and after electrofishing a closed population of brown trout Salmo trutta was determined by trapping in a small stream during 1994–1996. To assess the influence of movements on mark–recapture estimates and evaluate obtained estimates in relation to removal estimates, four different estimates (removal in summer, removal in autumn, conventional mark–recapture, and mark–recapture considering fish movements) were made for the closed brown trout population. The number of brown trout caught in upstream traps increased following electrofishing. For these fish, mean size was larger and recapture rates were lower compared with the general population in the stream. In all years the conventional mark–recapture method resulted in the highest population estimates and differed from the mark–recapture estimates in which fish movements were accounted for. Mark–recapture estimates accounting for movements did not differ from autumn removal estimates. The results emphasize the importance of minimizing disturbances in connection with studies of the population dynamics of stream fish.