All existing neutron sources suffer from disadvantages which would not be tolerated in modern megavoltage X-ray equipment. Experimental work with 30-60 MeV protons on beryllium and other elements has shown how these difficulties can to a large extent be overcome. Angular distributions and kerma rates as a function of proton energy are presented for various targets. Thin targets are found to be a better method than filtration for improving the penetration of the beam. A ²³⁷Np fission counter and a GM counter have been used to decompose stray radiation into its neutron and gamma-ray components. A model of a treatment head was found to attenuate the neutron component to less than 1% of its value of the useful beam.