Magnetomyography (MMG) can be used as a contactless modality to study the neuromuscular system. On the one hand, being contactless is a practical advantage as there is no need to prepare skin or attach electrodes as in electromyography (EMG). On the other hand, it is also a disadvantage because the magnetic field decays with increasing distance. However, the effect of sensor-to-source distance in MMG has not been systematically studied. Comparative in vivo and in silico experiments of the effect of sensor-to-source distance were performed. In vivo, muscle activity was recorded using simultaneous surface EMG and one triaxial optically pumped magnetometer (OPM). For the simulations, an established multiscale muscle model was used to predict how distance affects the signal-to-noise ratio (SNR) and the signal's spectral content. Given an environmental noise level of 0.5-1 pT root-mean-square (RMS) from 10 to 350 Hz, it was impossible to robustly detect muscle activity of one finger flexor muscle beyond a distance of two centimeters using OPM technology. In silico experiments showed a high SNR between 8 and 29 for MMG at 0.5 cm distance. Increasing the distance increases the MMG's median frequency content. The simulations uncovered that this is due to the effect of noise. For distances greater than two centimeters, measuring MMG of voluntary contractions in medium-sized muscles with current OPM technology and conventional magnetic shielding cannot be recommended.