Following spinal cord injury (SCI), the motor output flow to the limb(s) and sensory input to the brain is largely lost. While attempted movements with the paralysed and sensory deprived body part can still evoke signals in the sensorimotor system, this task-related 'net' brain activity of SCI patients differs substantially from healthy controls. Such reorganised and/or altered activity is thought to reflect abnormal processing. It is however possible that this altered net sensorimotor activity in SCI patients conceals preserved somatotopically-specific representations of the paralysed and sensory deprived body parts that could be exploited in a functionally meaningful manner (e.g. via neuroprosthetics). In this cross-sectional study, we investigated whether a functional connection between the periphery and the brain is necessary to maintain somatosensory representations. We used functional MRI and an (attempted) finger movement task to characterise the somatotopic hand layout in the primary somatosensory cortex and structural MRI to assess spared spinal tissue bridges. We tested 14 tetraplegic SCI patients (mean age, s.e.m.=55, 3.6; 1 female) who differed in terms of lesion completeness, retained sensorimotor functioning, and time since injury, as well as 18 healthy control participants (mean age, s.e.m.=56, 3.6 years; 1 female). Our results revealed somatotopically organised representations of patients' hands in which neighbouring clusters showed selectivity for neighbouring fingers in contralateral S1, qualitatively similar to those observed in healthy controls. To quantify whether these representations were normal in tetraplegic SCI patients we correlated each participant's intricate representational distance pattern across all fingers (revealed using representational similarity analysis) with a canonical inter-finger distance pattern obtained from an independent sample. The resulting hand representation typicality scores were not significantly different between patients and controls. This was even true when considering two individual patients with no sensory hand functioning, no hand motor functioning, and no spared spinal tissue bridges. However, a correlational analysis revealed that over years since SCI the hand representation typicality in primary somatosensory cortex deteriorates. We show that somatosensory representations can be maintained for several years following SCI even in the absence of perhiperhal inputs. Such preserved cortical hand representations could therefore be exploited in a functionally meaningful way by rehabilitation approaches that attempt to establish new functional connections between the hand and the brain after an SCI (e.g. through neuroprosthetics). However, time since injury may critically influence the somatotopic representations of SCI patients and might thereby impact the success of such rehabilitation approaches.