Except for amygdaloid lesions and cortical cerebellar ablations, selective destruction of virtually any part of the tel-, di-, mes-, or metencephalon was found to impede spatial reversal learning in the rat. (These data contrast sharply with those reported in recent studies on visual reversal learning in the rat.) Test trials given after learning each reversal problem disclosed significant (or marginally significant) deficits in retention only in those groups sustaining damage to the cingulate cortex, ventral hippocampus, rostral caudoputamen, lateral supraoptic hypothalamus, posterolateral hypothalamus, lateral thalamic complex, mediodorsal thalamus, ventromedial thalamus, parafascicular nucleus, substantia nigra, interpeduncular-central tegmental area, pontomesencephalic reticular formation, or pontine reticular formation. It is proposed that there may be at least three classes of learning (amnesic) disorders observed on any given task: (1) a learning impairment only, (2) a retention impairment only, and (3) both a learning and retention impairment.Huppert and Piercy (1979) have presented evidence suggesting that there may be at least two classes of learning (amnesic) disorders in braindamaged patients. In one, learning is impaired, but retention over time of what has been learned is unimpaired. The second is characterized by an impairment (or no impairment) in learning, but retention over time of what has been learned is impaired.