Abstract Dose-response curves for chromosome exchanges in Tradescantia paludosa microspores have been obtained using 0.7 MeV fast neutrons or 250 Kvp X-rays. The neutron curve shows clear evidence of saturation, and gives 2.18 for the value of N, the mean number of effective sites for exchange. When corrections are made for factors other than site limitation which will contribute to saturation, this value of N increases to 2.7. In contrast, the X-ray curve, although reaching a maximum exchange yield only 0.2 per cell below that attained for neutrons, shows no evidence of saturation. The difference may best interpreted in terms of differences in the numbers of exchange sites. The question remains as to whether this difference is intrinsic in the cells, or a reflection of qualitative differences between the radiations. In the former case, a mean number of 2.7 effective exchange sites per cell exists for neutrons, but considerably more for X-rays. This might indicate that the two radiations act at different sites. In the latter case two possibilities exist. 1. (a)The predominantly one-track nature of exchange formation with high LET radiations means that only a proportion of the actual sites present are available to neutrons. This proportion would consist of smaller sites where the probability of producing lesions in both threads with a single track of high LET is high. X-rays, although less efficient on a per track basis, have a large two-track component at higher doses, and would not be confined to the smaller sites, and would therfore have more sites available. 2. (b)The important factor may not be the number of sites available, but the existence of a progressive inactivation of sites (with consequent reduction in the number of exchanges produced) as the dose increases. If the efficiency of inactivation was greater at higher LET, the dose-response curve would saturate more quickly, and this would be interpreted as a lower number of effective sites.