||STRUCTURED ABSTRACT: Study Design. Multicenter, retrospective review.Objective. Assess rates, site variability, and risk factors for revision surgery (RS) following three-column osteotomy(3CO).Summary of Background Data. Complex spinal osteotomies, including 3CO, are being increasingly performed in the setting of adult spinal deformity (ASD) patients with sagittal plane deformity. 3CO procedures are associated with high complication and RS rates, but risk factors for complications and variability across centers for revision have not been well defined.Methods. The incidence and indications for RS in 335 ASD patients were analyzed. RS indications were classified as "mechanical"(MR: implant failure, pseudarthrosis, junctional failure, loss/lack of correction) or "non-mechanical"(NMR: neurologic deficit, infection, wound dehiscence, stenosis). Risks factors for RS were analyzed using generalized linear models.Results. Three-month and one-year RS incidences were 12.3% and 17.6%, respectively. Single-level 3CO (n = 311) had lower RS rates than multi-level 3CO(n = 24, 15.8% vs. 41.7%, p = 0.001). The 16.7% rate for single-level lumbar 3CO included 11.4% for MR and 5.7% for NMR. For all RS, 50% of MR and 78.6% of NMR occurred within three-months of the index surgery. There was significant variation in rates across sites(range = 6.3% to 31.9%, p = 0.001), however low- and high-volume sites had similar rates(18.2% vs. 16.2%, p = 0.503). Patients with MR were more likely to be sagittaly under-corrected at three-months(SVA = 7cm vs 3.2cm, p = 0.003). Patients with NMR had more caudal 3CO levels (L4 vs L3, p = 0.014) and larger 3CO bone resections than patients who did not(34° vs 24.5°, p = 0.003).Conclusions. 3CO procedures for ASD surgery can provide significant deformity correction and lead to marked improvement in function despite established complication and revision rates. This study shows that RS is associated with lower level osteotomy and higher residual SVA. There is significant variability in revision rates across sites independent of site volume, suggesting potential systems and practice variations that warrant further study.