Carp Cyprinus carpio L. populations of known composition as to size and descent were fished in drainable ponds in various ways to judge the validity of the mark-recapture method for estimation of population size. Six combinations of gear for capture and recapture were employed, including fyke-traps, seines and angling-rods. Satisfactory estimates could be obtained only when the selectivity of the gear for recapture differed from that used for marking. Some differences and changes in behaviour of the carp underlying observed biased estimates are examined.