The purpose of this study was to compare rod fracture (RF) rates among three types of rod constructs (RCs) following lumbar pedicle subtraction osteotomy (PSO) for adult spinal deformity (ASD).A retrospective review of consecutive patients with adult spinal deformity who were treated with lumbar PSO between 2007 and 2017 was performed. The minimum follow-up was 2 years. Three RCs were compared: standard (2 main rods), satellite (2 main rods with satellite rod), and nested (2 main rods and 2 short rods spanning osteotomy). Outcomes examined included RF rate, time to RF, pseudarthrosis, and reoperation. Multivariate analysis was used.A total of 141 patients were included 55 with standard, 23 with satellite, and 63 with nested RCs. The mean age was 65.2 years and 34.8% of patients were male. Radiographic preoperative and postoperative results were as follows: sagittal vertical axis (11.0 vs 3.9 cm), lumbar lordosis (28.5° vs 57.1°), pelvic tilt (30.6° vs 21.0°), pelvic incidence (61.5° vs 60.0°), distance between central sacral vertical line and C7 plumb line (2.2 vs 1.5 cm), and scoliosis (18.9° vs 11.3°). The average time to RF was 12.4 months. Overall RF, bilateral RF, pseudarthrosis, and reoperation rates were 22.7%, 5.0%, 20.6%, and 17.7%, respectively. Standard RCs had a significantly higher RF (36.4% vs 13.0% vs 14.3%, p = 0.008), bilateral RF (35.0% vs 0.0% vs 0.0%, p = 0.021), pseudarthrosis (34.5% vs 8.7% vs 12.7%, p = 0.004), and reoperation (30.9% vs 4.3% vs 11.1%, p = 0.004) rates. Satellite RCs (OR 0.21, p = 0.015), nested RCs (OR 0.24, p = 0.003), and bone morphogenetic protein-2 (OR 0.28, p = 0.005) were independently associated with lower odds of RF.Use of multiple rods in the satellite RC and nested RC groups was associated with lower rates of RF, pseudarthrosis, and reoperations following lumbar PSO. Bone morphogenetic protein-2 was associated with a reduction in RF rate as well.