Normally capable children were first taught to touch in sequence each of a set of five physically dissimilar stimuli (Sequence A). Another set of stimuli was then used to train sequence B. Next, direct training established conditional control of the production of the A sequence and its reversal: in the presence of one printed word, touching the stimuli in the order A1→A2→A3→A4→A5 was reinforced; in the presence of another word, touching the stimuli in the order A5→A4→A3→A2→A1 was reinforced. During probe sessions, the printed words also exercised conditional control over production of the B sequence and its reversal for five of six subjects, suggesting the formation of stimulus classes. Four of these five subjects also performed mixed sequences under conditional control of the words (e.g., A1→B2→A3→B4→A5 and its reversal), verifying that the stimuli which occupied the same position in each sequence were members of the same class.