The applicability of the EWF methodology to 3-point bend (SEB) specimens under conditions other than plane stress has been assessed experimentally. Different fracture conditions, pure plane strain and plane strain/plane stress transition, were obtained by varying the specimen thickness and testing temperature (20 and 80 degreesC). Post-mortem fracture surfaces appeared always completely stress-whitened, indicating ductile fracture. The load-line displacement plots are similar over a well-defined range of ligament lengths for which the application of the EWF methodology was in principle possible. Nevertheless, in experiments conducted at room temperature, crack growth was observed to initiate before maximum load and complete ligament yielding. This behaviour was confirmed through plastic collapse analyses. A critical ligament length was found, over which the total specific work of fracture was dominated by edge effects. Below this critical ligament length, EWF methodology was still applicable and it was possible to extrapolate reliable w(le) values.