Some determinants of the reinforcing and punishing properties of timeout were investigated in two experiments. Experiment I began as an attempt to reduce the frequency of tantrums in a 6-yr-old autistic girl by using timeout. Unexpectedly, the result was a substantial increase in the frequency of tantrums. Using a reversal design, subsequent manipulations showed that the opportunity to engage in self-stimulatory behavior during the timeout period was largely responsible for the increase in tantrums. Experiment II was initiated following the failure of timeout to reduce the spitting and self-injurious behavior of a 16-yr-old retarded boy. Using a multiple-baseline design, the nature of the timein environment was shown to be an important determinant of the effects of timeout. When the timein environment was "enriched", timeout was effective as a punisher. A conception of timeout in terms of the relative reinforcing properties of timein and timeout and their clinical implications are discussed.