Forty-eight adults learned intradimensional (ID) or extradimensional (ED) shifts in a total change paradigm under one of two conditions of dimensional dominance and one of two conditions of irrelevant dimension variability. Results showed a significant dominance main effect in initial training and significant Dominance by Shift and Variability by Shift interactions in shift training. When shifts were to a high-dominant dimension, there was no appreciable difference between ID and ED shifts. When shifts were to a low-dominant dimension, ID shift was easier than ED shift. The ID shift was learned faster than the ED shift when irrelevant dimensions in shift varied within trials. No reliable difference was found between ID and ED shifts when irrelevant dimensions varied between trials. These results support Esposito's (1975b) conclusion that the same principles guide shift behavior in children and adults.Esposito (l975a, b) suggested that for adults the apparent ease of intradimensional (ID) shifts over extradimensional (ED) shifts is partially a function of dimensional dominance. Several studies have shown that children find shifts to a high-dominant dimension easier than shifts to a low-dominant dimension (e.g., May & Fernandez, 1974). Esposito's work relating dominance to adult shift learning indicated that adults show the same type of discrimination shift behavior which is thought to characterize children's shift learning.In addition, Esposito (1975a) suggested that shift performance may be a function of irrelevant dimension variability. While this variable has received little attention, Dickerson, Wagner, and Campione (1970) found that 5-year-olds learned ID shifts faster than ED when the irrelevant dimensions varied within trials and that the ID-ED difference was negligible when the irrelevant dimensions varied between trials (constant within trials). Shepp and Gray (1971) and Zeaman and Denegre (1967) reported somewhat similar findings with kindergarten children and retardates, respectively. However, this variable has not been explicitly dealt with using adults as subjects.It has been suggested (e.g., Kendler & Kendler, 1962) that age differences in shift learning may be indicative of differences in basic learning processes between young children compared with adults. However, the similarity of results reported by Esposito (1975b) with adults and Dickerson et al. (I 970) with children suggest that dimensional dominance has a very similar effect upon the shift learning of both children and adults. One purpose of the present study was to replicate Esposito's fmdings with adults. A second purpose was to assess the generality of