The use of thoracic pedicle screws for the treatment of adolescent idiopathic scoliosis (AIS) has gained widespread popularity. However, the placement of pedicle screws in the deformed spine poses unique challenges, and surgeons experience a learning curve. The in vivo accuracy as determined by computed tomography (CT) of placement of thoracic pedicle screws in the deformed spine as a function of surgeon experience is unknown. We undertook a retrospective review to determine the effect of surgeon experience on the accuracy of thoracic pedicle screw placement in AIS. In 2005, we started to obtain routine postoperative CT scans on patients undergoing a spinal fusion. From a database of these patients, we selected AIS patients, who underwent a posterior spinal fusion. Fifteen consecutive patients for each of the following three groups stratified by attending surgeon experience were selected (N = 45): A) less than 20 cases of all pedicle screw constructs for AIS (surgeons \2 years of practice), B) 20-50 cases (surgeons 2-5 years of practice), and C) greater than 50 cases (surgeons greater than 5 years of practice). Intraoperative evaluation of all screws included probing of the pedicle screw tract, neurophysiologic monitoring, and fluoroscopic confirmation. A total of 856 thoracic pedicle screws were studied. Postoperative CT scans were evaluated by two spine surgeons and a consensus read established as follows: (1) In: intraosseous placement or B2-mm breach, (2) Out: [2-mm breach, either medial or lateral. Of the 856 screws, 104 demonstrated a [2-mm breach, for an overall rate of 12.1% (medial = 55, lateral = 49, P = 0.67). When the breach rates were stratified by surgeon experience, there was a trend toward decreased rate of breach for the most experienced surgeons, although this did not attain statistical significance (Group A: 12.7%, Group B: 12.9%, Group C: 10.8%, P = 0.58). However, the most experienced group (C) had a markedly decreased rate of medial breaches (3.5 vs. 7.4% and 8.4% for groups A and B, respectively, P \ 0.01). The breach rate for the concave periapical screws was not statistically different from the overall breach rate (13.0% vs. 12.1%, P = 0.93). In conclusion, the overall accuracy of placement of pedicle screws in the deformed spine was 87.9%, with no neurologic, vascular, or visceral complications. Meticulous technique allows spine surgeons with a range of surgical experience to accurately and safely place thoracic pedicle screws in the deformed spine. The most experienced surgeons demonstrated the lowest rate of medial breaches.