Unraveling the cellular and molecular mechanisms of spinal cord injury is fundamental for our possibility to develop successful therapeutic approaches. These approaches need to address the issues of the emergence of a non-permissive environment for axonal growth in the spinal cord, in combination with a failure of injured neurons to mount an effective regeneration program. Experimental in vivo models are of critical importance for exploring the potential clinical relevance of mechanistic findings and therapeutic innovations. However, the highly complex organization of the spinal cord, comprising multiple types of neurons, which form local neural networks, as well as short and long-ranging ascending or descending pathways, complicates detailed dissection of mechanistic processes, as well as identification/verification of therapeutic targets. Inducing different types of dorsal root injury at specific proximo-distal locations provide opportunities to distinguish key components underlying spinal cord regeneration failure. Crushing or cutting the dorsal root allows detailed analysis of the regeneration program of the sensory neurons, as well as of the glial response at the dorsal root-spinal cord interface without direct trauma to the spinal cord. At the same time, a lesion at this interface creates a localized injury of the spinal cord itself, but with an initial neuronal injury affecting only the axons of dorsal root ganglion neurons, and still a glial cell response closely resembling the one seen after direct spinal cord injury. In this review, we provide examples of previous research on dorsal root injury models and how these models can help future exploration of mechanisms and potential therapies for spinal cord injury repair.