The purpose of this study was to assess failure modes of knotless and knotted anchors in a Bankart repair model with the capsulolabral soft tissues intact. Previous reports used a model stripped of soft tissues. In 8 matched pairs of cadaver shoulders, a Bankart lesion was repaired arthroscopically using either 2 Bio-SutureTak anchors (Arthrex, Naples, Florida) or 2 Bioknotless anchors (Mitek, Westwood, Massachusetts). The shoulders were mounted with the repaired capsulolabral tissues attached to a custom sinusoidal clamp, and were tested in cyclic loading (20-80 N, 100 cycles, 0.5 mm/s) and then load to failure (1.25 mm/s). Cut-through at the suture-tissue interface (23/32 anchors) was more common than pullout at the anchor-bone interface (9/32) as a mode of failure (P = .02). Failure at the suture-tissue interface occurred in 10/16 knotted and 13/16 knotless anchors. Mean (SD) ultimate load of knotted vs knotless anchors was 125.3 (67.4) N and 96.9 (95.1) N, respectively. Mean (SD) stiffness of knotted vs knotless anchors was 20.9 (6.4) N/mm and 19.8 (8.6)N/mm, respectively. We concluded that both knotted and knotless anchors fail most often at the suture-tissue interface. The tested model with the capsulolabral tissues intact is distinct from previous models, which tested the anchor-bone interface only.