Amnestic mild cognitive impairment is characterised by deficits of episodic memory not severe enough to interfere with every day functioning. It is etiologically heterogeneous and variable in its course, but is considered as a transitional phase between intact cognitive function and dementia and may progress to Alzheimer’s dementia. Effective interventions to delay cognitive decline are needed. Negative changes in brain plasticity have been proposed as an important mechanism underlying cognitive impairments in the aging brain, involving noisy processing, reduced modulatory control, and disuse related to negative learning in the brain circuits involved in sensorimotor processes. We developed a brain-plasticity-based intervention targeting sensorimotor skills with the aim to enhance sensorimotor representations and thus positively impact memory performance. Forty-four persons with amnestic mild cognitive impairment participated in a double-blind randomised controlled clinical study. The training consisted of three months of home-based sensorimotor training in the visual, auditory, somatosensory and motor domain, enhanced by motivational features. It was compared to a validated cognitive training. We assessed the effects of these trainings on episodic memory performance pre- and post-intervention and at six-month follow-up. Using functional magnetic resonance imaging, we also measured the activation of hippocampal subfields (left dentate gyrus and left Cornu Ammonis area 2/3) during a mnemonic discrimination task involving objects and scenes, as well as changes in the somatotopic representation of the fingers of the trained right-hand. We also assessed changes in sensorimotor function. Both, the sensorimotor and the cognitive training preserved episodic memory function at post intervention and the six-month follow-up, compared to the decline typically reported in published studies. Pattern recognition memory declined significantly in both groups. In the experimental group, there was a significant reduction of the activation of the left dentate gyrus during the mnemonic discrimination task. In the somatosensory cortex, the experimental group showed a significant reduction of the relative distance of the thumb to all other digits during sensory stimulation, whereas the hand representation of the control group enlarged significantly. In addition, both measures, hippocampal activation and finger representation, were significantly related. Exploratory analyses involved dose-response effects of training, the role of biomarkers and single versus multi domain impairment. In this study we showed that a brain-plasticity-based sensorimotor training with motivational enhancement shows comparable effects as a cognitive intervention on the stabilization of episodic memory function. The sensorimotor training displayed unique effects on brain function by significantly reducing dysfunctional hippocampal activation during a mnemonic recognition task and by inducing more distinctive representations in the primary somatosensory cortex. The significant correlation of both brain processes suggests that the induction of improved sensorimotor processing may be a novel tool to impact on memory decline and its underlying processes.