Gergely et al. (2002) reported that children imitated a novel action - illuminating a light-box by using the forehead - after a delay significantly more often if the hands of the experimenter had been visible in comparison with if they had been covered. In an attempt to explore these findings we conducted two studies with a total N of 63 children. Both studies investigated deferred imitation of the action in two conditions, with the hands of the experimenter visible or covered, but the settings differed. Study 1 (n = 30; mean age = 16.6 months) was carried out in an unfamiliar environment (a laboratory setting) while Study 2 (n = 33; mean age = 13.3 months) was conducted in familiar surroundings (at home or at day care). The results showed that 50% of the children in Study 1 and 42.4% in Study 2 evidenced deferred imitation as compared to only 4.9% (n = 2) in the baseline condition. However, in none of the studies did the children use inferential processes when imitating, we detected no significant differences between the two conditions, hands visible or hands covered. The findings add to the validity of the head touch procedure as a measure of declarative-like memory processes in the pre-verbal child. At the same time the findings question the robustness of the concept rational imitation, it seems not as easy as expected to elicit a response based on rational inferential processes in this age group.