Background: Lumbar pedicle subtraction osteotomy (PSO) can significantly correct thoracolumbar kyphosis and relieve compensatory backward pelvic tilt (PT). However, hip involvement for some advanced ankylosing spondylitis (AS) patients can restrict the rotation of the pelvis around the femoral heads, which may affect the postoperative radiographic outcomes after lumbar PSO. Objective: To identify whether hip involvement negatively impacts the postoperative radiographic outcomes after lumbar PSO in AS patients with thoracolumbar kyphosis. Methods: Between March 2009 and June 2013, a total of 44 consecutive AS patients with thoracolumbar kyphosis who had undergone one-level lumbar PSO were retrospectively reviewed. All the patients had more than two years of follow-up. Hip involvement was evaluated based on the Bath Ankylosing Spondylitis Radiology Hip Index (BASRI-hip) and defined by a score of at least 2. All patients were divided into group A (patients without hip involvement) and group B (patients with hip involvement). Radiographical measurements included sagittal vertical axis (SVA), global kyphosis (GK), thoracic kyphosis (TK), local kyphosis (LK), lumbar lordosis (LL), pelvic incidence (PI), PT, and sacral slope (SS). The visual analogue pain scale for low back pain was also recorded. Results: Preoperative SVA and PT were not significantly different between group A and group B (SVA: 14.0 ± 6.3 cm vs. 14.7 ± 6.0 cm, P > 0.05; PT: 36.8° ± 7.1° vs. 37.3° ± 7.7°, P > 0.05), and both groups had similar magnitudes of kyphosis corrections (LK correction: 44.9° ± 4.9° vs. 44.2° ± 8.3°, P > 0.05). However, group B had significantly larger SVA and PT than group A (SVA: 7.6 ± 4.5 cm vs 3.5 ± 3.4 cm, P < 0.05; PT: 28.1° ± 8.6° vs. 19.0° ± 8.0°, P < 0.05) at the last follow-up. Conclusion: In AS patients with thoracolumbar kyphosis, hip involvement led to insufficient correction of SVA and PT after lumbar PSO, which negatively impacted postoperative radiographic outcomes. For these patients, additional osteotomies are recommended for satisfactory correction outcomes.