Bimodal size distributions have been commonly observed in Arctic char (Salvelinus alpinus). We document an example of such bimodality caused solely by biased gillnet sampling. The observed bimodality was a direct artefact of the sampling method resulting from an abrupt increase in gillnet catchability of fish larger in total length than between 25 and 30 cm. Mean gillnet selectivity (catchability) of char in the upper mode of the observed bimodal size distribution was about 40 times higher than the corresponding value for char in the observed lower mode. Fish of intermediate size, lacking in the gillnet samples, were present in the population and readily obtainable by electrofishing. The observed size difference in gillnet vulnerability is likely to result from behavioural changes following ontogenetic niche shifts.