In radiation physics, the biological effectiveness of radiation is generally estimated from the energy imparted to a sensitive volume of given size. However, measurements of this quantity are carried out almost exclusively with ionization-based techniques, by converting the ionization charge signal to energy imparted by means of a constant calibration factor, independent of particle type and energy. This procedure is reliable when the tissue-equivalent size of the sensitive volume is of the order of 1 μm and larger, since in this case the number of radiation-induced collisions is high and the distribution of energy transfer per single collision is masked by the multiple averaging. However, its reliability can be called into question when the size of the sensitive volume is decreased to the nanometre scale, as it is the case in experimental nanodosimetry. For this reason, the present study investigates the relationship between the two stochastic quantities energy imparted and ionization yield, calculated by Monte Carlo simulations in spherical sensitive volumes with diameter ranging from 100 nm down to 1 nm. The analysis was carried out for protons and carbon ions in the energy range from 1 to 100 MeV/u, crossing liquid-water sensitive spheres along their diameter. Simulations were done by means of the Geant4-DNA Monte Carlo code, with two different physics lists. A strong correlation between the energy imparted and the ionization yield was found, which comprises not only their mean values, but their entire stochastic distributions, even when the diameter of the sensitive volume is as small as 1 nm.