To investigate the role of inhibitory control in preschoolers' ability to switch sets, 3 conditions of the Dimensional Change Card Sort task ( Zelazo, Reznick, & Pinon, 1995) were tested. In Condition B (novel response options, standard stimuli) action inhibition was reduced, but the need for attentional inhibition was maintained. In Condition C (novel stimuli, standard response options) demands on both action and attentional inhibition were reduced. Performance in these was compared to that in the standard condition (A). Rule complexity was comparable across conditions. All 21 children who passed preswitch (average age 37 months) were tested on all postswitch conditions, order counterbalanced. Although reducing demands on action inhibition (Condition B) did not significantly improve performance, when demands on both action and attentional inhibition were reduced (Condition C) almost all children (95%) successfully switched sets (even children only 2(1)/(2)years old). Inadequate inhibition (of attention alone or both attention and action) appears sufficient to account for virtually all errors by preschoolers on this card sorting task.