Prior research has examined killing behavior using a paradigm in which participants believe (falsely) that they are killing bugs. This work suggests that killing behavior escalates. In the present study, we sought to replicate the basic escalation effect within-subjects. Further, in doing so, we controlled for experimenter “sanctioning” of killing that may have differed with key between-subjects manipulations in the prior research. To control for this possible confound, the present experiment held experimenter instructions constant and examined whether killing naturally escalated within-subjects across two 12-sec bug-killing tasks. Additionally, to verify that escalation is due to killing per se and not just physical practice of the procedure, we manipulated whether the procedure was described as real killing or simulated killing. Results showed that when participants thought they were killing bugs, the number of bugs put into the grinder increased from the first to the second killing task. No such escalation occurred when participants performed the procedure while knowing the killing was simulated. Thus, killing of bugs escalates and is not simply a consequence of perceived sanctioning of killing by an experimenter or simulated practice of the procedure.