PURPOSE: To identify which physical properties of nanoparticles are correlated with the survival fraction of cells exposed in vitro to low-energy protons in combination with nanoparticles. METHODS: The Geant4 simulation toolkit (version 10.3) was used to model nanoparticles of different sizes (5-50 nm) and materials (Ti, Zr, Hf, Ta, Au, Pt), with or without an organic capping ensuring biocompatibility and to irradiate them with 1.3 or 4 MeV protons and 5.3 MeV alpha particles. The spectra of secondary electrons inside and at the nanoparticle surface were computed, as well as electron yields, Auger and organic capping contribution, trapping in metal bulk and linear energy transfer profiles as a function of distance from the nanoparticle center. In a next step, an in silico cell model was designed and loaded with gold nanoparticles, according to experimental uptake values. Dose to the cell was evaluated macroscopically and microscopically in 100 × 100 × 100 nm³ voxels for different radiation qualities. RESULTS: The cell geometry showed that radiation enhancement is negligible for the gold concentration used and for any radiation quality. However, when the single nanoparticle geometry is considered, we observed a local LET in its vicinity considerably higher than for the water equivalent case (up to 5 keV/μm at the titanium nanoparticle surface compared to 2.5 keV/μm in the water case). The yield of secondary electrons per primary interaction with 1.3 MeV protons was found to be most favorable for titanium (1.54), platinum (1.44), and gold (1.32), although results for higher Z metals are probably underestimated due to the incomplete simulation of de-excitation cascade in outer shells. It was also found that the organic capping contributed mostly to the production of low-energy electrons, adding a spike of dose near the nanoparticle surface. Indeed, the yield for the coated gold nanoparticle increased to 1.53 when exposed to 1.3 MeV protons. Although most electrons are retained inside larger nanoparticles (50 nm), it was shown that their yield is comparable to smaller sizes and that the linear energy transfer profile is better. From a combination of ballistic and nanoparticle size factors, it was concluded that 10-nm gold nanoparticles were better inducers of additional cell killing than 5-nm gold nanoparticles, matching our previous in vitro study. CONCLUSIONS: Although effects from a physical standpoint are limited, the high linear energy transfer profile at the nanoparticle surface generates detrimental events in the cell, in particular ROS-induced damage and local heating.