Experimental studies have shown that diffuse axonal injury is usually induced by positive or negative acceleration mechanisms. In order to determine the reliability of axonal injury (AI) as a marker of this type of traumatic insult, we compared cases of trauma-induced focal cortical hemorrhage without dural involvement (n = 67) with cases of trauma-induced subdural bleeding without cortical hemorrhage (n = 26). Both groups exhibited a wide range of post-traumatic survival times. The injuries in the first group were caused mainly by direct impact to the head, those in the second by acceleration/deceleration mechanisms. The investigations were based primarily on immunohistochemical demonstration of antibodies targeted to beta-amyloid precursor protein (beta-APP) in the pons as a marker of AI and the results were assessed semiquantitatively. No significant differences were found between the two groups. In both groups AI was detected in 80-100% of cases with survival times of more than 3 h and two thirds of all positive cases showed pronounced positivity. Additional comparison of cases of brain death due to mechanical trauma (n = 14) with cases of brain death due to non-mechanical trauma (n = 18) also disclosed no significant intergroup differences. Finally, investigations of the pons in cases of non-traumatic death due to cerebral hypoxia/ischemia (n = 51) demonstrated AI with the same frequency as in the other groups, although the expression tended to be less pronounced. Our results confirm that beta-APP expression in the pons is a reliable indicator of AI but does not discriminate between injuries caused by traumatic strain or shearing mechanisms and secondary damage due to cerebral hypoxia/ischemia or edema. In the large majority of cases with prolonged post-traumatic survival, it can therefore be assumed that AI in the pons is the consequence of primary and/or secondary events or a combination of both, as is common in non-missile head injury survived for more than 90-120 min. Therefore, positive differentiation of the type of biomechanical event based on this criterion alone is not possible.