Background Microneurography is an established technique whereby a tungsten electrode is placed within a human peripheral nerve bundle, where single afferents can be distinguished. The electrode can be used to record the activity of a single unit, or by applying a short train of pulses at microampere levels, a localised sensation can be perceived, which has qualities that match the physiologically-defined unit. Single unit intra-neural microstimulation provides an extremely specific input to the brain, thought to be at the quantal level. Microneurography has been used for functional MRI, but it has not been tested in MEG, which provides excellent temporal resolution. Commercial systems have limitations due to: magnetic components (relays or other ferrites); potentially hazardous long cables; or long recovery when switching between recording and stimulating modes. Methods The key to the success of designing an instrument capable of fulfilling the requirements was using opto-coupled FET analogue switches to route connections from the electrodes to either the recording amplifier or the current stimulator. The amplifier comprised a low-noise instrumentation amplifier with a bandwidth of 1–10 kHz. The signals were digitised, filtered, recorded, and passed to an audio feed so that the microneurographer could hear the nerve signals whilst positioning the electrode. The instrumentation was opto-isolated for safe operation. A stimulus sequence could be generated using a fully analogue constant current driver and fed to the electrode. Measurement of the current delivered could be monitored and an impedance estimate of the electrode made. Noise and potential for interference was measured for both systems i.e. for both electrode and MEG recordings. Two separate MEG installations were tested for compatibility, a CTF and an Elekta system. Results Wide-band electrode recordings show a thermal noise commensurate with the electrode impedance. High quality recordings could be obtained from mechanoreceptive afferents originating in the hand. There was some electrical interference (50 Hz and lighting related) on the electrode recordings in the CTF, but this did not prevent successful microneurography. No interference was observed on the Elekta system. At stimulus levels usually used (<10 uA) in the median nerve, there was no detection of any stimulus artefact by the MEG sensors with conventional gradiometer cancellation. Beamformer techniques further reduced any possibility of interference. Discussion A MEG (and fMRI) compatible micro-stimulation system has been demonstrated. Brain activity in the somatosensory area can be observed mimicking single afferent unit activity. The present system adhered to safety requirements and provided good microneurographic use, for both recording and stimulation. This opens up opportunities for further work to explore the temporal and spatial dynamics of quantal tactile input in humans