Rotter's hypothesis that internals show more typical shifts in expectancies (increase in expectancies following success or decrease in expectancies following failure) was tested after some refinements in his formulation of the locus of control construct were made. 100 subjects were administered Internal-External, Perceived Competence, and Origin-Pawn scales. Then, 50 subjects were given the choice to take an anagrams task (free condition), and 50 subjects were required to participate (forced condition). Using a multiple regression procedure, the findings were that internal subjects tended to show more typical changes in expectancy than external ones in the free condition, but the results did not achieve statistical significance. Contrary to Rotter's hypothesis, in the forced condition external subjects showed more typical shifts in expectancy than internal subjects. The findings, in general, did not support Rotter's theory concerning the relationship of individual differences in locus of control beliefs to typical shifts in expectancies.