Neutron beams for therapy always contain some gamma ray contamination that varies with depth and with distance from the beam axis. The problem therefore arises how the varying gamma ray contribution should be accounted for in dose specification. Not infrequently a ``total effective dose'' DE is quoted that is equal to the neutron dose plus the gamma ray dose divided by a constant weight factor tau. On general biophysical considerations this appears to be not a valid approach since it must be assumed that tau decreases with increasing dose. The nature and the magnitude of this dose dependence is derived in the present article. Application of the results to actual doses per fraction and to factual gamma ray to neutron ratios demonstrates that the dose dependence of tau has, in fact, very minor influence on the numerical values of DE. Utilization of a constant value tau is therefore satisfactory in practice.