The renewal effect, which occurs when extinguished behaviour is tested outside of the extinction context, has been proposed as a model of contextual control that has relevance for the understanding of relapse following exposure therapy. Notably, there are multiple mechanisms by which the renewal effect can be explained. In two experiments, we used within-subjects designs in which participants learned to avoid a loud noise signalled by two discrete visual stimuli (CSs+), by pressing the space bar on the computer keyboard. Each CS+ was trained along with a CS- in a different context. During extinction, CS+ and CS- stimuli were presented in the alternative context from that of training, and participants were allowed to freely respond, but no loud noise was presented. Finally, all CSs were tested in both contexts, resulting in a within-subjects ABA vs ABB comparison. In both experiments, participants increased avoidance responses during training, and decreased responding during extinction, although Experiment 2 revealed less extinction. During test, responding was higher when CS+ stimuli were tested in the training (ABA) vs the extinction context (ABB), thus showing renewal of instrumental avoidance in humans. Additional analyses in Experiment 2 showed a remarkable similarity between avoidance responses and expectancy ratings. This study shows renewal of instrumental avoidance in humans, and the results suggest the operation of a modulatory role for the context in renewal, similar to occasion setting of extinction learning by the context.