An intriguing error has been observed in toddlers presented with a 3-location search task involving invisible displacements of an object, namely, the C-not-B task. In 3 experiments, the authors investigated the dynamics of the attentional focus process that is suspected to be involved in this task. In Experiment 1, 2.5-year-old children were tested on a new adaptation of the C-not-B task in which the opening of the experimenter's hand between cloths provided visual information about the correct localization of the toy. Children still emitted a strong response bias toward the last hiding place. In Experiment 2, 2.5-year-old children were tested on a new version of the task that was designed to investigate the role of the central location in the task. This 2nd experiment demonstrated that changing the hand's movement from A to C to B did not enable children to succeed in the task. In Experiment 3, 2.5-year-old children were tested in a situation that is analogous to the C-not-B with open hands task except for the fact that the experimenter dropped the toy under the 1st cloth in the path. Toddlers succeeded when the toy was hidden at Location A but not when it was hidden at Location B. Data indicate that attentional focus on the experimenter's hand motion is contingent on whether that stimulus is critical to performing the task. We argue that these findings provide a potential mechanism through which motor routines can be regulated in accordance with strategic intentions. (Contains 3 figures.)