<p>After an earthquake, the decision to repair or replace a damaged reinforced-concrete (RC) structure depends partly on crack width observations, which are used to classify damage levels. However, empirical crack width limits prescribed by current guidelines are coarsely defined and do not consider the effect of reinforcement configuration, axial load ratio or component size on the damage severity indicated by a crack measurement. This work adopts a simulation framework to assess which parameters affect crack widths at a severe, or “red-tag”, damage state. The so-called relative performance simulation framework quantifies the influence of a damaging earthquake on the collapse probability of a structure in a future earthquake and is implemented herein for seven squat RC shear walls from the experimental literature. The walls are modeled with a thoroughly validated macro-modeling approach and are simulated for 28 back-to-back incremental dynamic analyses to determine their damaged and undamaged collapse fragilities. By plotting the peak crack width during the damaging earthquake against the collapse capacity reduction due to the earthquake-induced damage, a “red-tag” crack width is obtained for each wall. Axial load ratio has a profound effect on red-tag crack width, which ranges from 0.7 mm to 5.5 mm for load ratios between 0 and 20%. It is therefore clear that crack width inspection triggers cannot be invariant, as is the case in existing guidelines.</p>