This study tests the effectiveness of an acceptance/defusion intervention in reducing experimentally induced generalized avoidance. After the formation of two 6‐member equivalence classes, 23 participants underwent differential conditioning with two elements from each class: A1 and B1 were paired with mild electric shock, whereas A2 and B2 were paired with earning points. Participants learned to produce avoidance and approach responses to these respective stimuli and subsequently showed transfer of functions to non‐directly conditioned equivalent stimuli from Class 1 (i.e., D1 and F1 evoked avoidance responses) and Class 2 (i.e., D2 and F2 evoked approach responses). Participants were then randomly assigned to either a motivational protocol (MOT) in which approaching previously avoided stimuli was given a general value, or to a defusion protocol (DEF) in which defusion (a component of Acceptance and Commitment Therapy) was trained while approaching previously avoided stimuli was connected to personally meaningful examples. A post‐hoc control group (CMOT) was conducted with 16 participants to control for differences in protocol length between the former two groups. All participants in the DEF group showed a complete suppression of avoidance responding in the presence of Class 1 stimuli (A1–F1 and additional novel stimuli in relation to them), as compared to 40% of participants in the MOT condition and 20% in the CMOT condition. The acceptance/defusion protocol eliminated experimentally induced avoidance responding even for stimuli that elicited autonomic fear responses.