The work is aimed at studying the reliability of proton-induced neutron yield cross sections on 12C–127I biological tissue isotopes using the GEANT4 Monte Carlo simulation package, which is commonly applied for dosimetry calculations in proton therapy. Because of the lack of experimental results on the neutron yield for most of the isotopes considered, the simulated cross sections are first compared with theoretical data acquired using the TALYS nuclear reaction package, and the partial reaction cross sections calculated in TALYS are in turn analyzed in comparison with available experimental results. The mismatch between theory and experiment allows one to suggest a significant uncertainty in the calculation of the neutron doses associated with the isotopes under consideration.