The task-switch cost is one of the most robust phenomena, but it can disappear after nogo trials where the actors decide not to respond to the target. According to the response-selection account, it is the occurrence of response selection that generates a task-switch cost on the following trial. The present study used a variety of selective go/nogo procedures to investigate whether response selection on nogo trials is followed by a switch cost. The first two experiments aimed to replicate previous studies in which go/nogo trials could be distinguished either by the target features or by the responses assigned to the target, but the results were mixed. The subsequent four experiments constrained the conditions so that the actors would need to select a specific response in order to decide whether or not they execute the response. In these experiments, the task-switch cost was consistently absent after selective nogo trials, even when response selection was required on these trials. These results contradict the hypothesis that response selection on nogo trials would be followed by a subsequent switch cost. The results are consistent with the proposal that a task-switch cost might have been established by response selection or other task-related process on nogo trials, but it is abolished because nogo signals interfered with the activated task-set.