Activity rate of Arctic char (Salvelinus alpinus) held in 90 m² littoral enclosures were estimated using bioenergetic (with consumption estimated using stable caesium, ¹³³Cs) and behavioural approaches (with fish movements quantified using video cameras). We found no statistically significant difference between values of activity rate obtained using the two approaches for three of the six experiments we performed. However, there was no relationship between estimates of activity rate obtained using the two approaches. Discrepancies may arise from the difficulty to meet assumptions regarding the temporal stability of the concentration of ¹³³Cs in fish diet and of the assimilation coefficient of this tracer. When fish remain in an area where their behaviour can be well described (e.g., enclosure, habitat patches of littoral zones, coral reefs), the behavioural approach appears more robust to estimate activity rate because it depends most on a variable that is easiest to estimate (the number of movements performed). When these conditions are not met (low fish densities or major fish migrations), a reliable assessment of the concentration and assimilation of ¹³³Cs in stomach contents appears critical to implement the bioenergetic approach based on this tracer.