The treatment of cancer using fast neutrons was first attempted from 1938 to 1942, only a few years after the identification of the particle in 1932. The radiobiological information which was available at that time was both inadequate and contradictory, and provided no definite rationale for using neutrons in preference to X-rays. The doses given were often too high, causing many patients to suffer severe late reactions. As a result, further attempts to use fast neutrons in radiotherapy were abandoned for nearly 30 years. Interest in the use of fast neutrons was stimulated again by the elucidation of the oxygen effect and the discovery that it was less for neutrons than for X-rays. Thus tumours containing hypoxic cells would be less protected against neutrons. Also the reduced repair of sublethal damage with neutrons provided at least a partial explanation of the miscalculation of dose in the early trial. This was confirmed by means of a series of experiments on pig skin, from which it was also concluded that late damage was not more severe after neutrons, compared with X-rays for a given degree of early damage. A new clinical trial began in 1966, and the results so far are encouraging. In order to relate radiotherapy experience with X-rays to neutrons, it is necessary to measure the relative biological effectiveness (RBE) of neutrons. This has been done for skin of man, pig, mouse and rat. Because of the smaller recovery from sublethal damage after neutrons, the RBE increases as the dose per fraction decreases, but the relationship between RBE and dose per fraction is the same for all four species. Similar information, but only for rodents, has been obtained for a variety of other normal tissues with both cyclotron-produced and monoenergetic 14 or 15 meV neutrons. Experiments with animal tumours have indicated that there might be a wide variation in RBE from tumour to tumour due both to the presence of hypoxic cells and to differences in their capacities to recover from sublethal damage after X-rays and neutrons. The largest series of experiments on one tumour shows that whereas certain fractionation techniques with X-rays may produce a poor tumour response for a given level of normal tissue damage, all the neutron regimes produced a similar, close to optimum result. There is no evidence from which to expect any special dangers from neutron irradiation, and their likely advantage is that they may provide a more reliable method of radiotherapy as well as sterilizing some tumours which are normally resistant to X-rays.