Abstract: Pigeons that had been trained with a food reward both to peck at and to step on a horizontal plate were allowed to observe a conspecific demonstrator pecking at or stepping on the plate before a test in which the observers were not rewarded for either pecking or stepping. In experiment 1, the demonstrators were not rewarded while being observed. In spite of this, the observers provided evidence of imitation: those that had observed pecking made a greater proportion of pecking responses on test than observers of stepping. In experiment 2, each observer was exposed to a pecking or a stepping conspecific on two occasions. On one occasion, the demonstrator received a food reward for each demonstrated response (continuous reinforcement condition), and on the other the demonstrator's responses were rewarded only rarely (variable interval condition). The observers provided equally strong evidence of imitation in each of these conditions; on test, they made proportionally more of the observed response both when the demonstrators had been richly rewarded and when they had been rarely rewarded. These results show that pigeons engage in `blind' imitation, that is, their imitative behaviour is not always guided by observational learning about response outcomes.