Rats given the option to choose between a signaled-shock condition, in which all signals were followed by shock, and one in which extra signals occurred showed no preference for either condition. In contrast, when all signals were followed by shock, rats showed a strong preference for a condition in which shock followed the signal by a fixed, rather than a variable, time duration. The latter result indicates that some types of information about aversive events may be reinforcing.Recent evidence indicates that preference for signaled over unsignaled inescapable shock is importantly controlled, not by the signal that precedes shock, but by stimuli accompanying the signal's absence (discriminable shock-free time or "safe" periods). However , there are conflicting data regarding the role of the signal as a predictor of shock (Arabian & Desiderato, 1975 ; Badia, Harsh , Coker, & Abbott, 1976). Part of this discrepancy could stem from the fact that stimuli which reliably predict shock contain at least two different types of information. The presence of the signal indicates that shock will occur, while the usual fixed temporal relationship between onset of the signal and onset of shock indicates something about when shock will occur. Badia et a1. (1976) manipulated the first source of information by varying the proportion of extra signals and found preference for the signaled over the unsignaled condition to be largely independent of this variable. Arabian and Desiderato (1975), on the other hand, manipulated the second source and found preference for a condition containing less, rather than more, accurate information about the time of shock delivery (but see Collier, 1977). The present research assessed the ability of the two sources of information described above to support preference behavior in a situation where the signal's absence always identified a period free from shock. Thus, any differences in observed responding could be attributed to the reinforcing effects of the signal-shock relationship rather than to stimuli associated with the signal's absence.