Despite growing knowledge about the mechanisms of fracture healing, non-union formation still represents a major complication in trauma and orthopedic surgery. Non-union models in mice gain increasing interest, because they allow investigating the molecular and cellular mechanisms of failed fracture healing. These models often use segmental defects to achieve non-union formation. Alternatively, failed fracture healing can be induced by transverse fractures with additional periosteal injury. The present study systematically compared the reliability of these two approaches to serve as non-union model. A 0.6 mm K-wire was inserted into the femora of CD-1 mice in a retrograde fashion and a closed transverse femoral fracture was created. Subsequently, the fracture site was exposed and the periosteum was cauterized. This approach was compared with a well-established non-union model involving the pin-clip fixation of a 1.8 mm segmental defect. The callus tissue was analyzed by means of radiography, biomechanics, histology and Western blotting. At 10 weeks after surgery 10 out of 12 femora (83.3 %) of the K-wire group showed a non-union formation. The pin-clip model resulted in 100 % non-union formation. The K-wire group showed increased bone formation, osteoclast activity and bending stiffness when compared to the group with pin-clip fixation. This was associated with a higher expression of bone formation markers. However, the number of CD31-positive microvessels was reduced in the K-wire group, indicating an impaired angiogenic capacity after periosteal cauterization. These findings suggest that the pin-clip model is more reliable for the study of non-union formation in mice. The K-wire model including periosteal injury by cauterization however, may be particularly applied in preclinical studies which explore the effects of damaged periosteum and reduced angiogenic capacity to trauma-induced fractures.