Herman's (1973) proposition that satisfaction and performance can only be related when performance is at least partially under the worker's control was tested. Subjects were hired to work under conditions of high or low situational control of performance. Two performance measures were used, only one of which was designed to be susceptible to the situational control manipulation. Satisfaction was then predicted by performance, situational control condition, and their cross-product. As expected, the interaction term was significant when the performance measure susceptible to situational control was used and nonsignificant when the nonsusceptible measure was used. It was concluded that the amount of performance variance which is individually rather than situationally controlled is an important moderator of the satisfaction-performance relationship.