Objective.To study the secondary neutrons generated by primary oxygen beams for cancer treatment and compare the results to those from primary protons, helium, and carbon ions. This information can provide useful insight into the positioning of neutron detectors in phantom for future experimental dose assessments.Approach.Mono-energetic oxygen beams and spread-out Bragg peaks were simulated using the Monte Carlo particle transport codes FLUktuierende KAskade, tool for particle simulation, and Monte Carlo N-Particle, with energies within the therapeutic range. The energy and angular distribution of the secondary neutrons were quantified.Main results.The secondary neutron spectra generated by primary oxygen beams present the same qualitative trend as for other primary ions. The energy distributions resemble continuous spectra with one peak in the thermal/epithermal region, and one other peak in the fast/relativistic region, with the most probable energy ranging from 94 up to 277 MeV and maximum energies exceeding 500 MeV. The angular distribution of the secondary neutrons is mainly downstream-directed for the fast/relativistic energies, whereas the thermal/epithermal neutrons present a more isotropic propagation. When comparing the four different primary ions, there is a significant increase in the most probable energy as well as the number of secondary neutrons per primary particle when increasing the mass of the primaries.Significance.Most previous studies have only presented results of secondary neutrons generated by primary proton beams. In this work, secondary neutrons generated by primary oxygen beams are presented, and the obtained energy and angular spectra are added as supplementary material. Furthermore, a comparison of the secondary neutron generation by the different primary ions is given, which can be used as the starting point for future studies on treatment plan comparison and secondary neutron dose optimisation. The distal penumbra after the maximum dose deposition appears to be a suitable location for in-phantom dose assessments.