Purpose: It is well-known that revision rates after primary knee arthroplasty vary widely. However, it is uncertain whether hospital revision rates are reliable indicators of general surgical quality as defined by patients. The SPARK study compared primary knee arthroplasty surgery at three high-volume hospitals whose revision rates differed for unknown reasons. Methods: This prospective observational study included primary knee arthroplasty patients (total, medial/lateral unicompartmental and patellofemoral) in two low-revision hospitals (Aarhus University Hospital and Aalborg University Hospital Farsø) and one high-revision hospital (Copenhagen University Hospital Herlev-Gentofte). Patients were followed from preoperatively (2016–17) to 1-year postoperatively with patient-reported outcome measures including Oxford Knee Score (OKS), EQ-5D-5L and Copenhagen Knee ROM (range of motion) Scale. The surgical outcomes were compared across hospitals for patients with comparable grades of radiographic knee osteoarthritis and preoperative OKS. Statistical comparisons (parametric and non-parametric) included all three hospitals. Results: 97% of the 1452 patients who provided baseline data (89% of those included and 56% of those operated) responded postoperatively (90% at 1 year). Hospitals’ utilization of unicompartmental knee arthroplasties differed (Aarhus 49%, Aalborg 14%, and Copenhagen 22%, p < 0.001). 28 patients had revision surgery during the first year (hospital independent, p = 0.1) and were subsequently excluded. 1-year OKS (39 ± 7) was independent of hospital (p = 0.1), even when adjusted for age, sex, Body Mass Index, baseline OKS and osteoarthritis grading. 15% of patients improved less than Minimal Important Change (8 OKS) (Aarhus 19%, Aalborg 13% and Copenhagen 14%, p = 0.051 unadjusted). Patients with comparable preoperative OKS or osteoarthritis grading had similar 1-year results across hospitals (OKS and willingness to repeat surgery, p ≥ 0.087) except for the 64 patients with Kellgren–Lawrence grade-4 (Aarhus 4–6 OKS points lower). 86% of patients were satisfied, and 92% were “willing to repeat surgery”, independent of hospital (p ≥ 0.1). Hospital revision rates differences diminished during the study period. Conclusions: Patients in hospitals with a history of differing revision rates had comparable patient-reported outcomes 1 year after primary knee arthroplasty, supporting that surgical quality should not be evaluated by revision rates alone. Future studies should explore if revision rate variations may depend as much on revision thresholds and indications as on outcomes of primary surgery. Level of evidence: Level II (Prospective cohort study).