The accurate diagnosis of "osteomyelitis" is difficult and is often delayed. However, early and radical therapy of osteomyelitis is essential. The osteomyelitis diagnosis score (ODS) was developed to predict the probability of osteomyelitis. The aim of our study was to validate the ODS and to investigate its practicability in daily routine. The ODS is based on five diagnostic procedures: 1. clinical history/risk factors, 2. clinical examination/laboratory results, 3. diagnostic imaging, 4. microbiology and 5. histopathology. Each diagnostic procedure includes numerous individual findings, which are rated with 1-6 points, depending on their relevance. If the sum of the five diagnostic criteria is ≥ 17 points, the diagnosis "osteomyelitis" can be viewed as safe, between 8-17 points as probable and between 2-7 points as possible. This retrospective study included 100 patients with non-union of the tibia (2002-2010). The patients were classified into two groups: septic non-union of the tibia (experimental intervention; gold standard: positive detection of bacteria and/or positive histology) and aseptic non-union of the tibia (control intervention; no detection of bacteria and/or histology). Epidemiological data, the score's total number of points and the number of points of the score's five diagnostic procedures were analysed. 71 patients exhibited aseptic non-union of the tibia, 29 patients septic non-union. Patients with septic non-union obtained a mean of 20.8 points, and 24 at least 18 points; the diagnosis "osteomyelitis" is then presumed to be certain. Patients with aseptic non-union obtained a mean of 11.3 points, and only 3/71 patients received > 17 points. Both groups obtained the majority of points in the diagnostic procedure "clinical history". The difference between the two groups is highly significant (p < 0.001). The score's sensitivity is 82.8 %, with a specificity of 95.8 %. The ODS was proved to be a valid score. Patients with septic non-union were identified, even if bacteria were not detected. However, the use of the ODS is demanding, as there are 104 individual findings. Many of these individual findings were negative in all patients. It would be desirable to optimise ODS, by reducing the number of queried parameters, without reduction sensitivity.