One-many paired-associate lists were constructed by pairing several responses to the same stimulus. When those pairs having the same stimulus were presented consecutively during study trials, better recall was obtained on test trials than otherwise. Presumably, the former condition allowed subjects to form vertical associations among those response members paired to the same stimulus as well as horizontalassociations between the stimulus and response members of pairs. Another finding that more intralist intrusion errors were obtained in the one-many than in the one-one paired-associate learning also supports the above reasoning.