To report the anatomic and functional outcomes of Descemet's stripping automated endothelial keratoplasty (DSAEK) in 32 eyes after failed penetrating keratoplasty (PK).This retrospective, single-center observational study was performed on 32 eyes of 26 patients with failed PK grafts who underwent DSAEK between June 2009 and June 2013 at Clermont-Ferrand University Medical Center. Primary outcomes measured were: graft survival rates, functional improvement (visual acuity), and complications.Mean follow-up was 16.45 months [6-36]. Graft survival at 12 months was 76.1 % [68.1-82.3]. PK grafts cleared in 75 % of cases. Eight primary DSAEKs detached and/or decentered early (prior to day 8, average 6.22 days): 3 underwent repeat DSAEK with a new graft, and 3 underwent conversion to PK. Results were available for 26 patients at 6 months. Mean preoperative best-corrected visual acuity (BCVA) in logMAR increased from 1.48 (counting fingers at 2 meters) to 0.996 logMAR (1/10) at 6 months. BCVA improved in 17 patients (68 %). BCVA improved an average of 5.2 lines (P=0.0006). Eight patients had a large number of comorbid conditions that limited final visual acuity.Anatomic outcomes are similar to other studies. Final average BCVA was limited by severe comorbid conditions and performance of DSAEK solely for comfort. Many surgical strategies have been described to decrease graft dislocation and primary graft failure. Adapting graft trephine diameter to the host cornea as measured by AS-OCT, using a nomogram based on posterior corneal curvature obtained on AS-OCT, may be an interesting approach.DSAEK allows for quick visual recovery by preserving the anterior corneal curvature. It is an essential option for restoring corneal clarity in PK failure due to endothelial decompensation. A longer follow-up would allow a comparison of graft survival for the two techniques in this context.