In this third and final part, the Spine Study Group (AG WS) of the German Trauma Association (DGU) presents the follow-up (NU) data of its second, prospective, internet-based multicenter study (MCS II) for the treatment of thoracic and lumbar spinal injuries including 865 patients from 8 trauma centers. Part I described in detail the epidemiologic data of the patient collective and the subgroups, whereas part II analyzed the different methods of treatment and radiologic findings. The study period covered the years 2002 to 2006 including a 30-month follow-up period from 01.01.2004 until 31.05.2006. Follow-up data of 638 (74%) patients were collected with a new internet-based database system and analyzed. Results in part III will be presented on the basis of the same characteristic treatment subgroups (OP, KONS, PLASTIE) and surgical treatment subgroups (Dorsal, Ventral, Kombi) in consideration of the level of injury (thoracic spine, thoracolumbar junction, lumbar spine). After the initial treatment and discharge from hospital, the average duration of subsequent inpatient rehabilitation was 4 weeks, which lasted significantly longer in patients with persistent neurologic deficits (mean 10.9 weeks) or polytraumatized patients (mean 8.6 weeks). Following rehabilitation on an inpatient basis, subsequent outpatient rehabilitation lasted on average 4 months. Physical therapy was administered significantly longer to patients with neurologic deficits (mean 8.7 months) or type C injuries (mean 8.6 months). The level of injury had no influence of the duration of the inpatient or outpatient rehabilitation. A total of 382 (72.2%) patients who were either operated from posterior approach only or in a combined postero-anterior approach had an implant removal after an average 12 months. During the follow-up period 56 (8.8%) patients with complications were registered and of these 18 (2.8%) had to have surgical revision. The most common complications reported were infection, loss of correction, or implant-associated complications. Clinical data showed a 2.9 higher relative risk for smokers compared to non-smokers to suffer from wound healing problems. The neurologic status of 81 (60.4%) out of 134 patients with neurologic deficits at the time of injury improved until follow-up. Neurologic deterioration was documented in 8 (1.3%) cases. Complete neurologic deficits after injury to the thoracic spine improved in 9% of the cases, whereas 59% of the cases with complete neurologic deficit improved after injury to the thoracolumbar junction. The surgical approach (posterior or combined postero-anterior) had no significant influence on neurological results at follow-up. Patient age, sex and neurologic deficits showed a statistically significant influence (p<0.05) on the fingertip-floor distance (FBA) at follow-up. Patient back function improved during the follow-up period. More than 2 years after the time of injury 32.2% of the patients had no complaints with respect to back function. The relative frequency of patients with unrestrained back function was greater after posterior surgery (24.2%), than anterior surgery (13.8%), or combined surgery (17.3%) (p=0.005; chi(2)-test). At follow-up there were no statistically significant differences of unrestrained back function between different levels of injury (thoracic spine 17.4%, TL junction 22.5% and lumbar spine 13.6%). The relative frequency of patients with injury to the thoracolumbar junction who reported "no complaints from the anterior approach" at follow-up, was calculated to be 55.6% after open versus 63.8% after endoscopic approaches with no significant differences. Of the patients 56.3% reported no donor site morbidity following iliac crest bone harvesting. The VAS spine score at follow-up was calculated within different treatment subgroups: OP 58.4 points, KONS 59.8 points, and PLASTIE 59.7 points. Statistically significant differences of the VAS spine score between posterior (64.9 points) versus combined surgery (47.8 points) were only verified at the level of injury of the thoracic spine (p=0.004). The relative frequency of patients regaining at least 80% of the initial score level was OP (posterior 60.4%, anterior 61.1%, combined 51.4%), 52.9% KONS and 67.6% PLASTIE. After surgery the mean period of incapacity from work was 4 months. Patients with a sedentary occupation before the time of injury were fully reintegrated into work in 71.1% of the cases. Patients with a physical occupation were fully reintegrated in 38.9% of the cases at follow-up. At follow-up 87 (31.2%) patients after posterior and 50 (20.1%) after combined surgery had no restrictions to their recreational activities (p=0.001). Treatment subgroups PLASTIE and KONS show a similar radiological result at follow-up with a bisegmental kyphotic deformity (GDW) of -9 degrees and -8.5 degrees, respectively. With all operative methods it was possible to correct or partly correct the posttraumatic kyphotic deformity. Until follow-up there was a loss of correction depending on the surgical approach and level of injury. Combined postero-anterior stabilization gave statistically significant better radiological results with less kyphotic deformity (-3.8 degrees) than posterior stabilization alone (-6.1 degrees) (p=0.005; ANOVA). Thus combined surgery was superior in its capability to restore spinal alignment within the observational period. At follow-up the use of titanium vertebral body replacement implants (cages) to reconstruct and support the anterior column showed significantly better radiological results with less kyphotic deformity and loss of correction (GDW 0.3 degrees) than the use of iliac bone strut grafts (-3.7 degrees ) (p<0.001). Neither additional anterior plates nor the combination of anterior plates with a cage or bone graft had a statistically significant influence on the kyphotic deformity measured at follow-up. A matched-pair analysis of anterior surgery alone versus combined surgery for the treatment of compression fractures (type A) at the thoracolumbar junction showed a significantly greater intraoperative blood loss but better radiological results in terms of monosegmental and bisegmental kyphotic deformity after combined surgery (p<0.05). A matched-pair analysis of treatment results between non-operative and operative treatment for burst fractures (type A3.1-2) showed a period of inability to work (6 months) which was twice as long for the non-operative treatment group. At the same time significantly better radiological results at follow-up were achieved after operative treatment of these fractures (p<0.05).