We report a new model of transient axonal stretch injury involving pressurized fluid deflection of bundles of axons, resulting in a transient 1-6% increase in original axon length to investigate the slow progression of axonal alterations that are characteristic of diffuse axonal injury (DAI). We found no discernable difference in axon bundle morphology or cytoskeletal neurofilament protein arrangement between unstretched and stretched axonal bundles at 24 h post-injury. However, by 48 h post-injury, there was a stereotypical response of stretched axons involving characteristic neurofilament alterations that bear similarities to in vivo neuronal responses associated with DAI that have been reported previously. For instance, neurofilament protein immunoreactivity (SMI-312) was increased in axons contained within 51% of all injured axon bundles at 48 h compared to surrounding unstretched axon bundles, suggestive of neurofilament compaction. Furthermore, axonal bundle derangement occurred in 25% of injured axon bundles, with individual fibres segregating from each other and becoming undulating and wavy. By 72 h post-stretch, 70% of injured axon bundles underwent secondary axotomy, becoming completely severed at the site of initial stretch injury. While these results suggest a temporal series of stereotypical responses of axons to injury, we were able to distinguish very clear differences between mildly (100-103% increase in original axonal length) injured and strongly injured (106%+) axons. For instance, mildly injured axons developed increased neurofilament immunoreactivtity (SMI-312) within 48 h, and the marked development of ring-like neurofilament immunoreactive structures within axonal bundles, which were rarely axotomized. Conversely, at more severe strain levels increased neurofilament immunoreactivity was less apparent, while axons often became distorted and disorganised within axonal bundles and eventually became completely disconnected. Almost no ring-like neurofilament structures were observed in these severely injured axonal bundles. This suggests that axons do not respond in a stereotypical manner to a transient stretch insult, and indeed that variable degrees of stretch injury activate different responses within axons, with dramatically different outcomes. Hence, it is possible that the cytoskeletal characteristics that we have used in this study may be useful parameters for discriminating between mildly and severely injured axons following TBI.