Abstract.Juvenile hatchery‐reared Atlantic salmon,Salmo salarL., were offered choices of pelletted or four types of wild prey (chironomid larvae, ephemeropteran nymphsEcdyanurusandBaetisand tnchopteran larvaeHydropsyche) of the same particle widths, in a test flume Preference for wild foods increased over 16 trials (1·5 h), and was greatest forEcdyonurtaandHydropsyche, which had prey width fish fork length (PFR) ratios of 0·019 and 0·022 respectively, Pelletted food was rejected sooner whenEcdyonurusnymphs were offered as the alternative food than when chironomid larvae were (P<0·05). Pret'erence (P<0·05) forEcdyonuruswas achieved after 11 trials and for chironomids after 15. The time required to capture the first prey item was two to three times as long for pellets as for wild prey and did not change with experience for any prey type. Mean capture distance did not change with experience. It is concluded that acceptability of wild prey should not limit performance of pellet‐fed salmon released into