Three separate effects on refuge use by signal crayfish Pacifastacus leniusculus and Atlantic salmon Salmo salar were examined: (1) the effect on Atlantic salmon of an addition of signal crayfish (doubling the total number of animals), (2) the effect on signal crayfish of an addition of Atlantic salmon and (3) intraspecific compared with interspecific competition, compared by holding total density of animals constant and varying the proportion of signal crayfish and Atlantic salmon in trials. Observations were made during winter, when both species are nocturnal. The proportion of Atlantic salmon sheltering was significantly lower in the presence than in the absence of signal crayfish when the interspecific treatment (Atlantic salmon plus signal crayfish) effected a doubling in density compared to the intraspecific treatment (Atlantic salmon alone). The proportion of signal crayfish sheltering was independent of the presence of Atlantic salmon. When total density was constant, the proportion of Atlantic salmon sheltering was significantly higher in intraspecific (52·8%) than interspecific trials (27·3%). Atlantic salmon out of shelter during the day in winter are believed to be very vulnerable to predators and the capacity for fish to share shelters with one another is known to be very low. Therefore, competition from crayfish for winter shelters may lead to detrimental effects on Atlantic salmon populations.