Selectivity of recreational angling on fish behaviour was studied by examining whether capture order or lure type (natural v. artificial bait) in ice-fishing could explain behavioural variation among perch Perca fluviatilis individuals. It was also tested if individually assessed personality predicts fish behaviour in groups, in the presence of natural predators. Perca fluviatilis showed individually repeatable behaviour both in individual and in group tests. Capture order, capture method, condition factor or past growth rate did not explain variation in individual behaviour. Individually determined boldness as well as fish size, however, were positively associated with first entrance to the predator zone (i.e. initial risk taking) in group behaviour tests. Individually determined boldness also explained long-term activity and total time spent in the vicinity of predators in the group. These findings suggest that individual and laboratory-based boldness tests predict boldness of P. fluviatilis in also ecologically relevant conditions, i.e. in shoals and in the presence of natural predators. The present results, however, also indicate that the above-mentioned two angling methods may not be selective for certain behavioural types in comparison to each other.