A 52.5‐cm‐long steel collimator for 15‐MeV neutron therapy beams has been evaluated in terms of the distribution of the collimator‐scattered neutrons in the field and penumbra, the distribution in the collimator of the neutron scattering sites, and the energy of the neutrons scattered to one position in the penumbra, in order to assess the need for multiple tapers on such collimators. The associated particle technique was used to eliminate background neutrons from the measurement. We conclude that because collimator‐scattered neutrons constitute only a small part of the neutron flux in the field and most of the scattered neutrons have energies greater than 12 MeV, double‐taper collimators are not necessary.