Objective: Plexus injury results in lifelong suffering of flaccid paralysis, sensory loss, and intractable pain. For this clinical problem, regenerative medicine concepts, such as cell replacement for restoring dorsal root ganglion (DRG) function, set high expectations. However, it is completely unclear which DRG cell types are affected by plexus injury. Methods: We investigated the cellular composition of human DRG in a clinically characterized cohort of patients with plexus injury. Avulsed DRG of 13 patients were collected during reconstructive nerve surgery. Then, we analyzed the cellular composition of the DRG with a human-adapted objective deep learning-based analysis of large-scale microscopy images. Results: Surprisingly, in about half of the patients, the injury-affected DRG no longer contained DRG cells. The complete entity of neurons, satellite glial cells, and microglia was lost and replaced by mesodermal/connective tissue. In the other half of patients, the cellular entity of the DRG was well preserved. We found no loss of neurons, no gliosis, and macrophages close to single sensory neuron/satellite glial cell entities. Patients with "neuronal preservation" had less pain than patients with "neuronal loss". Interpretation: The findings classify plexus injury patients in two categories: type I (neuronal preservation) and type II (neuronal loss). We call for early, post-accidental interventions to protect the entire DRG and improved MRI diagnostics to detect "neuronal loss". Regenerative medicine to restore DRG function will need at least two translational directions: reafferentation of existing DRG units for type I injuries; or replacement of the entire DRG structure for type II patients.