SIMPLE reaction time tasks usually contain procedures designed to reduce or eliminate the frequency with which subjects anticipate the onset of the stimulus. These measures include varying the interval or foreperiod between the warning stimulus and the subsequent stimulus, and introducing a number of “catch trials” on which the stimulus does not follow the warning. Obviously, one effect of including catch trials is to lower the overall probability that the stimulus will follow the warning stimulus. In view of the known importance of stimulus probability in choice reaction time situations, it is perhaps surprising that, although there have been several studies (for example, refs. 3–5) of the effects of changes in the range and variability of foreperiods, there appears to have been only a single study in which the effect of changing the frequency of catch trials has received systematic investigation. Drazin2 introduced catch trials to produce three levels of probability of occurrence of the stimulus (s): P(s) = 1.0, 0.9 and 0.5. These levels were allowed to interact with mean fore-period, so that Drazin's results regarding the effects of P(s) on reaction time are limited to the demonstration of (1) an inverse relationship between P(s) and reaction time, and (2) an interaction between P(s) and the reaction time–foreperiod relationship. The study reported here was designed to examine the relationship of the P(s) and reaction time over a wide range of P(s) values in a sequential, simple reaction time situation.