Grazing on two red tide dinoflagellates, the potentially toxic Karenia mikimotoi and the non-toxic Gyrodinium instriatum, was examined in two species of marine copepods, Pseudocalanus elongatus and Temora longicornis. Both copepods cleared K. mikimotoi at rates that were a little lower but comparable to those at which they cleared the slightly larger G. instriatum when the two dinoflagellates were offered separately. However, when feeding on mixtures of the two prey species, the clearance rates on K. mikimotoi were substantially reduced in both copepods while their clearances of G. instiatum remained unaltered, suggesting active prey selection. Video observations of individual prey capture and feeding events showed prey rejection frequencies (caught and then released cells) that did not differ between mixed and mono-specific diets. This suggests that the selection between prey cells occurs prior to capture and that it is based on remote characterization of the cells.