Aims. The aim of this study was to compare the osseous reactions elicited by all-suture, polyetheretherketone (PEEK), and two different biodegradable anchors used during rotator cuff repair. Patients and Methods. Transosseous-equivalent rotator cuff repair was performed in 73 patients. The patients were divided into two groups, in both of which two different medial-row anchors were used. In group 1, anchor A comprised 30% β-tricalcium phosphate (TCP) + 70% fast-absorbing poly lactic-co-glycolic acid copolymer (85% polylactic acid enantiomers + 15% polyglycolic acid) and anchor B comprised all-sutures. In group 2, anchor C comprised 23% micro β-TCP + 77% polylactic acid enantiomers and anchor D comprised PEEK polymer. There were 37 patients in group 1 and 36 patients in group 2. The presence and severity of fluid collection around anchors and healing of the rotator cuff were assessed using MRI scans, approximately one year postoperatively. The severity of the collection was graded as 0 (no perianchor fluid signal), 1 (minimal perianchor fluid), 2 (local collection of fluid), 3 (fluid collection around the whole length of the anchor but of a diameter less than twice the anchor diameter), or 4 (fluid collection around the whole length of the anchor and of a diameter greater than twice the anchor diameter). Results. A perianchor fluid signal was seen in three patients (8.1%) with anchor A, four (10.8%) with anchor B, 15 (41.7%) with anchor C, and 15 (41.7%) with anchor D. The severity of the collection around anchor was grade 2:1:0:0 for anchor A, grade 2:2:0:0 for anchor B, grade 12:2:0:1 for anchor C, and grade 11:3:0:1 for anchor D (grade 1:2:3:4, respectively). The prevalence and severity of fluid formation was not significantly different between anchors A and B, and anchors C and D. However, on intergroup analysis, there were significant differences for the prevalence and severity of fluid formation between anchors in group 1 and group 2. The prevalence of failure to heal was not significant in group 1 (seven, 18.9%) or group 2 (nine, 25.0%). There was no relationship between the presence of perianchor fluid and each type of anchor, and the integrity of the rotator cuff repair, in either group. Conclusion. Despite the nonabsorbable nature of all-suture and PEEK anchors, all-suture anchors produced less osseous reaction after rotator cuff repair. In deciding which kind of anchor to use, consideration should be given not only to the strength of the initial fixation, but also to the postoperative biological reaction. Cite this article: Bone Joint J 2019;101-B:1506–1511